Develop Reference

Advantages of Higher order functions in Python

To implement prettified xml, I have written following code
def prettify_by_response(response, prettify_func):
root = ET.fromstring(response.content)
return prettify_func(root)
def prettify_by_str(xml_str, prettify_func):
root = ET.fromstring(xml_str)
return prettify_func(root)
def make_pretty_xml(root):
rough_string = ET.tostring(root, "utf-8")
reparsed = minidom.parseString(rough_string)
xml = reparsed.toprettyxml(indent="\t")
return xml
def prettify(response):
if isinstance(response, str) or isinstance(response, bytes):
return prettify_by_str(response, make_pretty_xml)
else:
return prettify_by_response(response, make_pretty_xml)
In prettify_by_response and prettify_by_str functions, I pass function make_pretty_xml as an argument
Instead of passing function as an argument, I can simply call that function.e.g
def prettify_by_str(xml_str, prettify_func):
root = ET.fromstring(xml_str)
return make_pretty_xml(root)
One of the advantage that passing function as an argument to these function over calling that function directly is, this function is not tightly couple to make_pretty_xml function.
What would be other advantages or Am I adding additional complexity?

This seem very open to biased answers I'll try to be impartial but I can't make any promise.
First, high order functions are functions that receive, and/or return functions. The advantages are questionable, I'll try to enumerate the usage of HoF and elucidate the goods and bads of each one
Callbacks
Callbacks came as a solution to blocking calls. I need B to happens after A so I call something that blocks on A and then calls B. This naturally leads to questions like, Hmm, my system wastes a lot of time waiting for things to happen. What if instead of waiting I can get what I need to be done passed as an argument. As anything new in technology that wasn't scaled yet seems a good idea until is scaled.
Callbacks are very common on the event system. If you every code in javascript you know what I'm talking about.
Algorithm abstraction
Some designs, mostly the behavioral ones can make use of HoF to choose some algorithm at runtime. You can have a high-level algorithm that receives functions that deal with low-level stuff. This lead to more abstraction code reuse and portable code. Here, portable means that you can write code to deal with new low levels without changing the high-level ones. This is not related to HoF but can make use of them for great help.
Attaching behavior to another function
The idea here is taking a function as an argument and returning a function that does exactly what the argument function does, plus, some attached behavior. And this is where (I think) HoF really shines.
Python decorators are a perfect example. They take a function as an argument and return another function. This function is attached to the same identifier of the first function
#foo
def bar(*args):
...
is the same of
def bar(*args):
...
bar = foo(bar)
Now, reflect on this code
from functools import lru_cache
#lru_cache(maxsize=None)
def fib(n):
if n < 2:
return n
return fib(n-1) + fib(n-2)
fib is just a Fibonacci function. It calculates the Fibonacci number up to n. Now lru_cache attach a new behavior, of caching results for already previously calculated values. The logic inside fib function is not tainted by LRU cache logic. What a beautiful piece of abstraction we have here.
Applicative style programming or point-free programming
The idea here is to remove variables, or points and combining function applications to express algorithms. I'm sure there are lots of people better than me in this subject wandering SO.
As a side note, this is not a very common style in python.
for i in it:
func(i)
from functools import partial
mapped_it = map(func, it)
In the second example, we removed the i variable. This is common in the parsing world. As another side node, map function is lazy in python, so the second example doesn't have effect until if you iterate over mapped_it
Your case
In your case, you are returning the value of the callback call. In fact, you don't need the callback, you can simply line up the calls as you did, and for this case you don't need HoF.
I hope this helps, and that somebody can show better examples of applicative style :)
Regards

Python and function within function recursion

Say I have something like this sample code.
def foo(n):
def bar():
return -1
if n = 0:
return 0
else:
return foo(n+bar())
I assume it creates a new instance of bar every time foo is recursively called. However this seems like something that could be optimized in python (or any other language) but I was unable to find anything stating if it has been optimized for that.
The reasoning I have bar defined in foo is I'm trying to hide bar from the user and python's _bar() or __bar() of "please sir don't use this dear user" is annoying me as I was trained in non scripting languages.

def is an executable statement in Python (and so is class). A new function object for bar is created each time foo() is invoked, but it's pretty trivial cost. It just retrieves the compiled code object, and wraps it in a new function object. People stress over this waaaay too much ;-) In general, it's necessary to do this in order for closures to work right, and to capture appropriate default arguments. You're right that in many cases this could be optimized a bit, but the CPython implementation doesn't bother.

Creating a function object from a string

Question: Is there a way to make a function object in python using strings?
Info: I'm working on a project which I store data in a sqlite3 server backend. nothing to crazy about that. a DAL class is very commonly done through code generation because the code is so incredibly mundane. But that gave me an idea. In python when a attribute is not found, if you define the function __getattr__ it will call that before it errors. so the way I figure it, through a parser and a logic tree I could dynamically generate the code I need on its first call, then save the function object as a local attrib. for example:
DAL.getAll()
#getAll() not found, call __getattr__
DAL.__getattr__(self,attrib)#in this case attrib = getAll
##parser logic magic takes place here and I end up with a string for a new function
##convert string to function
DAL.getAll = newFunc
return newFunc
I've tried the compile function, but exec, and eval are far from satisfactory in terms of being able to accomplish this kind of feat. I need something that will allow multiple lines of function. Is there another way to do this besides those to that doesn't involve writing the it to disk? Again I'm trying to make a function object dynamically.
P.S.: Yes, I know this has horrible security and stability problems. yes, I know this is a horribly in-efficient way of doing this. do I care? no. this is a proof of concept. "Can python do this? Can it dynamically create a function object?" is what I want to know, not some superior alternative. (though feel free to tack on superior alternatives after you've answered the question at hand)

The following puts the symbols that you define in your string in the dictionary d:
d = {}
exec "def f(x): return x" in d
Now d['f'] is a function object. If you want to use variables from your program in the code in your string, you can send this via d:
d = {'a':7}
exec "def f(x): return x + a" in d
Now d['f'] is a function object that is dynamically bound to d['a']. When you change d['a'], you change the output of d['f']().

can't you do something like this?
>>> def func_builder(name):
... def f():
... # multiline code here, using name, and using the logic you have
... return name
... return f
...
>>> func_builder("ciao")()
'ciao'
basically, assemble a real function instead of assembling a string and then trying to compile that into a function.

If it is simply proof on concept then eval and exec are fine, you can also do this with pickle strings, yaml strings and anything else you decide to write a constructor for.

Is there a way to convert code to a string and vice versa in Python?

The original question was:
Is there a way to declare macros in Python as they are declared in C:
#define OBJWITHSIZE(_x) (sizeof _x)/(sizeof _x[0])
Here's what I'm trying to find out:
Is there a way to avoid code duplication in Python?
In one part of a program I'm writing, I have a function:
def replaceProgramFilesPath(filenameBr):
def getProgramFilesPath():
import os
return os.environ.get("PROGRAMFILES") + chr(92)
return filenameBr.replace("<ProgramFilesPath>",getProgramFilesPath() )
In another part, I've got this code embedded in a string that will later be
output to a python file that will itself be run:
"""
def replaceProgramFilesPath(filenameBr):
def getProgramFilesPath():
import os
return os.environ.get("PROGRAMFILES") + chr(92)
return filenameBr.replace("<ProgramFilesPath>",getProgramFilesPath() )
"""
How can I build a "macro" that will avoid this duplication?

Answering the new question.
In your first python file (called, for example, first.py):
import os
def replaceProgramFilesPath(filenameBr):
new_path = os.environ.get("PROGRAMFILES") + chr(92)
return filenameBr.replace("<ProgramFilesPath>", new_path)
In the second python file (called, for example, second.py):
from first import replaceProgramFilesPath
# now replaceProgramFilesPath can be used in this script.
Note that first.py will need to be in python's search path for modules or the same directory as second.py for you to be able to do the import in second.py.

No, Python does not support preprocessor macros like C. Your example isn't something you would need to do in Python though; you might consider providing a relevant example so people can suggest a Pythonic way to express what you need.

While there does seem to be a library for python preprocessing called pypp, I am not entirely familiar with it. There really is no preprocessing capability for python built-in. Python code is translated into byte-code, there are no intermediate steps. If you are a beginner in python I would recommend avoiding pypp entirely.
The closest equivalent of macros might be to define a global function. The python equivalent to your C style macro might be:
import sys
OBJWITHSIZE = lambda x: sys.getsizeof(x) / sys.getsizeof(x[0])
aList = [1, 2, 4, 5]
size = OBJWITHSIZE(aList)
print str(size)
Note that you would rarely ever need to get the size of a python object as all allocation and deletion are handled for you in python unless you are doing something quite strange.
Instead of using a lambda function you could also do this:
import sys
def getSize(x):
return sys.getsizeof(x) / sys.getsizeof(x[0])
OBJWITHSIZE = getSize
aList = [1, 2, 4, 5]
size = OBJWITHSIZE(aList)
print str(size)
Which is essentially the same.
As it has been previously mentioned, your example macro is redundant in python because you could simply write:
aList = [1, 2, 4, 5]
size = len(aList)
print str(size)

This is not supported at the language level. In Python, you'd usually use a normal function or a normal variable where you might use a #define in C.

Generally speaking if you want to convert string to python code, use eval. You rarely need eval in Python. There's a module somewhere in the standard library that can tell you a bit about an objects code (doesn't work in the interp), I've never used it directly. You can find stuff on comp.lang.python that explains it.
As to 'C' macros which seem to be the real focus of your question.
clears throat DO NOT USE C MACROS IN PYTHON CODE.
If all you want is a C macro, use the C pre processor to pre process your scripts. Duh.
If you want #include, it's called import.
If you want #define, use an immutable object. Think const int foo=1; instead of #define foo 1. Some objects are immutable, like tuples. You can write a function that makes a variable sufficiently immutable. Search the web for an example. I rather like static classes for some cases like that.
If you want FOO(x, y) ... code ...; learn how to use functions and classes.
Most uses of a 'CPP' macro in Python, can be accomplished by writing a function. You may wish to get a book on higher order functions, in order to handle more complex cases. I personally like a book called Higher Order Perl (HOP), and although it is not Python based, most of the book covers language independent ideas -- and those ideas should be required learning for every programmer.
For all intents and purposes the only use of the C Pre Processor that you need in Python, that isn't quite provided out of box, is the ability to #define constants, which is often the wrong thing to do, even in C and C++.
Now implementing lisp macros in python, in a smart way and actually needing them... clears throat and sweeps under rug.

Well, for the brave, there's Metapython:
http://code.google.com/p/metapython/wiki/Tutorial
For instance, the following MetaPython code:
$for i in range(3):
print $i
will expand to the following Python code:
print 0
print 1
print 2
But if you have just started with Python, you probably won't need it. Just keep practicing the usual dynamic features (duck typing, callable objects, decorators, generators...) and you won't feel any need for C-style macros.

You can write this into the second file instead of replicating the code string
"""
from firstFile import replaceProgramFilesPath
"""

Decorators in Ruby (migrating from Python)

I'm spending today learning Ruby from a Python perspective. One thing I have completely failed to grapple with is an equivalent of decorators. To pare things down I'm trying to replicate a trivial Python decorator:
#! /usr/bin/env python
import math
def document(f):
def wrap(x):
print "I am going to square", x
f(x)
return wrap
#document
def square(x):
print math.pow(x, 2)
square(5)
Running this gives me:
I am going to square 5
25.0
So, I want to create a function square(x), but decorate it so it alerts me as to what it's going to square before it does it. Let's get rid of the sugar to make it more basic:
...
def square(x):
print math.pow(x, 2)
square = document(square)
...
So, how do I replicate this in Ruby? Here's my first attempt:
#! /usr/bin/env ruby
def document(f)
def wrap(x)
puts "I am going to square", x
f(x)
end
return wrap
end
def square(x)
puts x**2
end
square = document(square)
square(5)
Running this generates:
./ruby_decorate.rb:8:in `document': wrong number of arguments (0 for 1) (ArgumentError)
from ./ruby_decorate.rb:15:in `<main>'
Which I guess is because parentheses aren't mandatory and it's taking my return wrap as an attempt to return wrap(). I know of no way to refer to a function without calling it.
I've tried various other things, but nothing gets me far.

Here's another approach that eliminates the problem with conflicts between names of aliased methods (NOTE my other solution using modules for decoration is a good alternative too as it also avoids conflicts):
module Documenter
def document(func_name)
old_method = instance_method(func_name)
define_method(func_name) do |*args|
puts "about to call #{func_name}(#{args.join(', ')})"
old_method.bind(self).call(*args)
end
end
end
The above code works because the old_method local variable is kept alive in the new 'hello' method by fact of define_method block being a closure.

Ok, time for my attempt at an answer. I'm aiming here specifically at Pythoneers trying to reorganize their brains. Here's some heavily documented code that (approximately) does what I was originally trying to do:
Decorating instance methods
#! /usr/bin/env ruby
# First, understand that decoration is not 'built in'. You have to make
# your class aware of the concept of decoration. Let's make a module for this.
module Documenter
def document(func_name) # This is the function that will DO the decoration: given a function, it'll extend it to have 'documentation' functionality.
new_name_for_old_function = "#{func_name}_old".to_sym # We extend the old function by 'replacing' it - but to do that, we need to preserve the old one so we can still call it from the snazzy new function.
alias_method(new_name_for_old_function, func_name) # This function, alias_method(), does what it says on the tin - allows us to call either function name to do the same thing. So now we have TWO references to the OLD crappy function. Note that alias_method is NOT a built-in function, but is a method of Class - that's one reason we're doing this from a module.
define_method(func_name) do |*args| # Here we're writing a new method with the name func_name. Yes, that means we're REPLACING the old method.
puts "about to call #{func_name}(#{args.join(', ')})" # ... do whatever extended functionality you want here ...
send(new_name_for_old_function, *args) # This is the same as `self.send`. `self` here is an instance of your extended class. As we had TWO references to the original method, we still have one left over, so we can call it here.
end
end
end
class Squarer # Drop any idea of doing things outside of classes. Your method to decorate has to be in a class/instance rather than floating globally, because the afore-used functions alias_method and define_method are not global.
extend Documenter # We have to give our class the ability to document its functions. Note we EXTEND, not INCLUDE - this gives Squarer, which is an INSTANCE of Class, the class method document() - we would use `include` if we wanted to give INSTANCES of Squarer the method `document`. <http://blog.jayfields.com/2006/05/ruby-extend-and-include.html>
def square(x) # Define our crappy undocumented function.
puts x**2
end
document(:square) # this is the same as `self.document`. `self` here is the CLASS. Because we EXTENDED it, we have access to `document` from the class rather than an instance. `square()` is now jazzed up for every instance of Squarer.
def cube(x) # Yes, the Squarer class has got a bit to big for its boots
puts x**3
end
document(:cube)
end
# Now you can play with squarers all day long, blissfully unaware of its ability to `document` itself.
squarer = Squarer.new
squarer.square(5)
squarer.cube(5)
Still confused? I wouldn't be surprised; this has taken me almost a whole DAY. Some other things you should know:
The first thing, which is CRUCIAL, is to read this: http://www.softiesonrails.com/2007/8/15/ruby-101-methods-and-messages. When you call 'foo' in Ruby, what you're actually doing is sending a message to its owner: "please call your method 'foo'". You just can't get a direct hold on functions in Ruby in the way you can in Python; they're slippery and elusive. You can only see them as though shadows on a cave wall; you can only reference them through strings/symbols that happen to be their name. Try and think of every method call 'object.foo(args)' you do in Ruby as the equivalent of this in Python: 'object.getattribute('foo')(args)'.
Stop writing any function/method definitions outside of modules/classes.
Accept from the get-go that this learning experience is going to be brain-melting, and take your time. If Ruby isn't making sense, punch a wall, go make a cup of coffee, or take a night's sleep.
Decorating class methods
The above code decorates instance methods. What if you want to decorate methods directly on the class? If you read http://www.rubyfleebie.com/understanding-class-methods-in-ruby, you find there are three methods for creating class methods -- but only one of them works for us here.
That is the anonymous class << self technique. Let's do the above but so we can call square() and cube() without instantiating it:
class Squarer
class << self # class methods go in here
extend Documenter
def square(x)
puts x**2
end
document(:square)
def cube(x)
puts x**3
end
document(:cube)
end
end
Squarer.square(5)
Squarer.cube(5)
Have fun!

Python-like decorators can be implemented in Ruby. I won't try to explain and give examples, because Yehuda Katz has already published a good blog post about decorators DSL in Ruby, so I highly recommend to read it:
Python Decorators in Ruby
Source code and tests
UPDATE: I've got a couple of vote downs on this one, so let me explain further.
alias_method (and alias_method_chain) is NOT exactly the same concept as a decorator. It is just a way to re-define method implementation without using inheritance (so client code won't notice a difference, still using the same method call). It could be useful. But also it could be error-prone. Anyone who used Gettext library for Ruby probably noticed that its ActiveRecord integration has been broken with each Rails major upgrade, because aliased version has been following the semantics of an old method.
The purpose of a decorator in general is NOT to change the internals of any given method and still be able to call the original one from a modified version, but to enhance the function behavior. The "entry/exit" use case, which is somewhat close to alias_method_chain, is only a simple demonstration. Another, more useful kind of a decorator could be #login_required, which checks authorization, and only runs the function if authorization was successful, or #trace(arg1, arg2, arg3), which could perform a set of tracing procedures (and be called with different arguments for different methods decoration).

What you might achieve with decorators in Python, you achieve with blocks in Ruby. (I cannot believe how many answers are on this page, without a single yield statement!)
def wrap(x)
puts "I am going to square #{x}"
yield x
end
def square(x)
x**2
end
>> wrap(2) { |x| square(x) }
=> I am going to square 2
=> 4
The concept is similar. With the decorator in Python, you're essentially passing the function "square" to be called from within "wrap". With the block in Ruby, I'm passing not the function itself, but a block of code inside of which the function is invoked, and that block of code is executed within the context of "wrap", where the yield statement is.
Unlike with decorators, the Ruby block being passed doesn't need a function to be part of it. The above could have been simply:
def wrap(x)
puts "I am going to square #{x}"
yield x
end
>> wrap(4) { |x| x**2 }
=> I am going to square 4
=> 16

This is a slightly unusual question, but interesting. I'd first strongly recommend that you don't try and directly transfer your Python knowledge to Ruby - it's better to learn the idioms of Ruby and apply them directly, rather than try to transfer Python directly. I've used both languages a lot, and they're both best when following their own rules and conventions.
Having said all that, here's some nifty code that you can use.
def with_document func_name, *args
puts "about to call #{func_name}(#{args.to_s[1...-1]})"
method(func_name).call *args
end
def square x
puts x**2
end
def multiply a, b
puts a*b
end
with_document :square, 5
with_document :multiply, 5, 3
this produces
about to call square(5)
25
about to call multiply(5, 3)
15
which I'm sure you'll agree does the job.

IMO mooware has the best answer so far and it is the cleanest, simplest and most idiomatic. However he is making use of 'alias_method_chain' which is part of Rails, and not pure Ruby. Here is a rewrite using pure Ruby:
class Foo
def square(x)
puts x**2
end
alias_method :orig_square, :square
def square(x)
puts "I am going to square #{x}"
orig_square(x)
end
end
You can also accomplish the same thing using modules instead:
module Decorator
def square(x)
puts "I am going to square #{x}"
super
end
end
class Foo
def square(x)
puts x**2
end
end
# let's create an instance
foo = Foo.new
# let's decorate the 'square' method on the instance
foo.extend Decorator
# let's invoke the new decorated method
foo.square(5) #=> "I am going to square 5"
#=> 25

Michael Fairley demonstrated this at RailsConf 2012. Code is available here on Github. Simple usage examples:
class Math
extend MethodDecorators
+Memoized
def fib(n)
if n <= 1
1
else
fib(n - 1) * fib(n - 2)
end
end
end
# or using an instance of a Decorator to pass options
class ExternalService
extend MethodDecorators
+Retry.new(3)
def request
...
end
end

Your guess is right.
You best use alias to bind the original method to another name, and then define the new one to print something and call the old one. If you need to do this repeatedly, you can make a method that does this for any method (I had an example once, but cannot find it now).
PS: your code does not define a function within a function but another function on the same object (yes, this is an undocument feature of Ruby)
class A
def m
def n
end
end
end
defines both m and n on A.
NB: the way to refer to a function would be
A.method(:m)

Okay, found my code again that does decorators in Ruby. It uses alias to bind the original method to another name, and then define the new one to print something and call the old one. All this is done using eval, such that it can be reused like decorators in Python.
module Document
def document(symbol)
self.send :class_eval, """
alias :#{symbol}_old :#{symbol}
def #{symbol} *args
puts 'going to #{symbol} '+args.join(', ')
#{symbol}_old *args
end"""
end
end
class A
extend Document
def square(n)
puts n * n
end
def multiply(a,b)
puts a * b
end
document :square
document :multiply
end
a = A.new
a.square 5
a.multiply 3,4
Edit: here the same with a block (no string manipulation pain)
module Document
def document(symbol)
self.class_eval do
symbol_old = "#{symbol}_old".to_sym
alias_method symbol_old, symbol
define_method symbol do |*args|
puts "going to #{symbol} "+args.join(', ')
self.send symbol_old, *args
end
end
end
end

I believe the corresponding Ruby idiom would be the alias method chain, which is heavily used by Rails. This article also considers it as the Ruby-style decorator.
For your example it should look like this:
class Foo
def square(x)
puts x**2
end
def square_with_wrap(x)
puts "I am going to square", x
square_without_wrap(x)
end
alias_method_chain :square, :wrap
end
The alias_method_chain call renames square to square_without_wrap and makes square an alias for square_with_wrap.
I believe Ruby 1.8 doesn't have this method built-in, so you would have to copy it from Rails, but 1.9 should include it.
My Ruby-Skills have gotten a bit rusty, so I'm sorry if the code doesn't actually work, but I'm sure it demonstrates the concept.

In Ruby you can mimic Python's syntax for decorators like this:
def document
decorate_next_def {|name, to_decorate|
print "I am going to square", x
to_decorate
}
end
document
def square(x)
print math.pow(x, 2)
end
Though you need some lib for that. I've written here how to implement such functionality (when I was trying to find there something in Rython that is missing in Ruby).