PIC 16: Iterators and Generators

Transcription

1 PIC 16: Iterators and Generators Assigned 10/9/2018. To be completed before lecture 10/12/2018. Intended Learning Outcomes. By the end of this preparatory assignment, students should be able to: implement iterator behavior in custom classes so they can be looped through using the same convenient syntax as built-in containers; use generator functions to create iterators easily, and use generator expressions to create simple generators. Tasks. Read 9.9. The concepts may seem confusing and convoluted at first, but the explanation in the tutorial is actually very clear and concise, so try to follow it line by line. If that doesn t work, here are some (verbose) thoughts that might help. We re learning to write the blueprint for our own container objects, that is, objects that contain multiple elements that we can access individually and iterate over in a for loop. Python lists, tuples, sets, and dictionaries are all built-in container objects. Now we are trying to create our own. Our custom container might have a built-in container object as an instance variable. For instance, the tutorial s Reverse class has an instance variable data that is already a container. In this case, we just need to write some code that tells a for loop how to iterate over that instance variable. The simplest sort of container object will have its own next method that, when called, returns to the for loop the next element in the container. When there are no more elements in the container, it raises a StopIteration exception (see 8.4) instead of returning an element. The for loop terminates when it gets this exception. In general, however, the container object does not need to have its own next method. Instead, it may assign the job of picking the next element to a separate object, called an iterator. An iterator is any object that defines a suitable next method. When an iterator object s next method is invoked, the method should return the next element of some collection - whatever that may mean. How the next method is written defines the order in which the elements of a collection are iterated over in a for loop. Your collection appoints an iterator by defining an iter method that returns an instance of an iterator object. 1

2 If the collection has its own next method, the collection s iter method can return self; the container will serve as its own iterator. Note that in the tutorial s example, the Reverse class is both the container and the iterator object. But in general, the iterator can be a separate object from the container. If you are understand all this, you might ask Wouldn t it be simpler if containers were just required to have their own next method that the for loop would call? What is the use of first getting an iterator object from an iter method (which could just be the container object itself) and then invoking that object s next method? You ll see in the assignment. The following code might help clarrify the fact that for loops use iterators in Python. def myfor(container, dothisonelement): iterator = iter(container) keepgoing = True while keepgoing: element = iterator.next() dothisonelement(element) keepgoing = False def p(x): print x, l = [0,1,2,3] s = "hello" t = (6,6,6,"the number of the beast") myfor(l,p); myfor(s,p); myfor(t,p) Read and understand the following code (which continues on the next page). It will not run as written because self.i = 0 is commented out twice. Uncomment each individually, and observe how the code runs. Explain it to yourself. class ThreeElementContainer: def init (self, a = 0, b = 0, c = 0): #self.i = 0 self.a = a self.b = b self.c = c def iter (self): #self.i = 0 print "iter called" return self 2

3 #continued... def next(self): print "next called" if self.i == 0: el = self.a elif self.i == 1: el = self.b elif self.i == 2: el = self.c else: print "raised StopIteration" raise StopIteration self.i += 1 return el def str (self): return "[" + str(self.a) + ", " + str(self.b) + ", " + str(self.c) + "]" c = ThreeElementContainer(5,10,15) print c, "\n" Read Here s the example above using a generator... class ThreeElementContainer: def init (self, a = 0, b = 0, c = 0): self.a = a self.b = b self.c = c def iter (self): print "iter called" return self.generator() def generator(self): print "1st yield" yield self.a print "2nd yield" yield self.b print "3rd yield" yield self.c print "post 3rd yield" 3

4 def str (self): return "[" + str(self.a) + ", " + str(self.b) + ", " + str(self.c) + "]" c = ThreeElementContainer(5,10,15) print c, "\n" Do two for loops work correctly? Write a generator every_other(data) that yields every other element of the data. The test code: for char in every_other("supercalifragilisticexpialidocious"): print char, should print: s p r a i r g l s i e p a i o i u. Read Note the first example might come in handy in the MathVector assignment. The examples don t make clear exactly how generator expressions work because they immediately apply a function to the generator expressions. Maybe the following will help... r = 1-1.0/8888 N = genexp1 = (r**i for i in xrange(0,n)) genexp2 = (r**i for i in xrange(0,n)) genexp3 = (r**i for i in xrange(0,n)) genexp4 = (r**i for i in xrange(0,n)) print genexp1, genexp2 print genexp3, genexp4 for i in xrange(10): print genexp1.next() == r**i, print "\n" print sum(genexp2) print genexp3.next() print sum(genexp3) print sum(genexp4) genexp4.next() 4

5 xrange is iterable exactly like range, but it avoids creating a list: l = range(3) i = iter(l) print l, type(l) print type(i) print i.next(), i.next(), i.next() i.next() x = xrange(3) j = iter(x) print x, type(x) print type(j) print j.next(), j.next(), j.next() j.next() Create the same generator every_other (as above) in a single line using a lambda function and a generator expression. Especially if you are totally confused, you might find these other references useful: Remove the next method and index instance variable from the Reverse class from the tutorial. Write a new class, ReverseIterator, that serves as an iterator for the Reverse class. Modify the iter method of the Reverse class accordingly so that the test code from the tutorial (for char in Reverse('spam'): print char) still works. 5