from module import *.
-5.3. Defining Classes
-Python is fully object-oriented: you can define your own classes, inherit from your own or built-in classes, and instantiate the - classes you've defined. -
Defining a class in Python is simple. As with functions, there is no separate interface definition. Just define the class and start coding. A Python class starts with the reserved word class, followed by the class name. Technically, that's all that's required, since a class doesn't need to inherit from any other
-class.
-
Example 5.3. The Simplest Python Class
-class Loaf: ①
- pass ② ③-
-
- The name of this class is
Loaf, and it doesn't inherit from any other class. Class names are usually capitalized,EachWordLikeThis, but this is only a convention, not a requirement. - - This class doesn't define any methods or attributes, but syntactically, there needs to be something in the definition, so
- you use
pass. This is a Python reserved word that just means “move along, nothing to see here”. It's a statement that does nothing, and it's a good placeholder when you're stubbing out functions or classes. - - You probably guessed this, but everything in a class is indented, just like the code within a function,
ifstatement,forloop, and so forth. The first thing not indented is not in the class. --
The passstatement in Python is like an empty set of braces ({}) in Java or C. -Of course, realistically, most classes will be inherited from other classes, and they will define their own class methods -and attributes. But as you've just seen, there is nothing that a class absolutely must have, other than a name. In particular, -C++ programmers may find it odd that Python classes don't have explicit constructors and destructors. Python classes do have something similar to a constructor: the
__init__method. + + +[classes stuff was here] + + +Example 5.4. Defining the
FileInfoClassfrom UserDict import UserDict diff --git a/dip3.css b/dip3.css index 8694be4..1dba446 100644 --- a/dip3.css +++ b/dip3.css @@ -101,7 +101,7 @@ h1,h2{letter-spacing:-1px} h1,h1 code{font-size:xx-large} h2,h2 code{font-size:x-large} h3,h3 code{font-size:large} -h1{border-bottom:4px double;width:100%;margin:1em 0} +h1{border-bottom:4px double;width:100%;margin:1em 0;text-shadow:gainsboro 1px 1px 1px} h1:before{content:"Chapter " counter(h1) ". "} h1{counter-reset:h2} h2:before{counter-increment:h2;content:counter(h1) "." counter(h2) ". "} diff --git a/examples/fibonacci2.py b/examples/fibonacci2.py index 6a3fa31..8aeef15 100644 --- a/examples/fibonacci2.py +++ b/examples/fibonacci2.py @@ -1,6 +1,6 @@ """Fibonacci iterator""" -class fib: +class Fib: def __init__(self, max): self.max = max diff --git a/iterators-and-generators.html b/generators.html similarity index 66% rename from iterators-and-generators.html rename to generators.html index 7fa6971..d6da5ea 100644 --- a/iterators-and-generators.html +++ b/generators.html @@ -1,7 +1,7 @@ -Iterators & generators - Dive into Python 3 +Generators - Dive into Python 3 + + + +You are here: Home ‣ Dive Into Python 3 ‣ +
Iterators
++
+❝ East is East, and West is West, and never the twain shall meet. ❞
— Rudyard Kipling ++
Diving In
+Generators are really just a special case of iterators. A function that
yields values is a nice, compact way of building an iterator without building an iterator. Remember the Fibonacci generator? Here it is as a built-from-scratch iterator: + +
+ +class Fib: + def __init__(self, max): + self.max = max + + def __iter__(self): + self.a, self.b = 0, 1 + return self + + def __next__(self): + fib = self.a + if fib > self.max: + raise StopIteration + self.a, self.b = self.b, self.a + self.b + return fibLet's take that one line at a time. + +
+ +class Fib:class? What's a class? + +Defining Classes
+ +Python is fully object-oriented: you can define your own classes, inherit from your own or built-in classes, and instantiate the classes you've defined. + +
Defining a class in Python is simple. As with functions, there is no separate interface definition. Just define the class and start coding. A Python class starts with the reserved word
class, followed by the class name. Technically, that's all that's required, since a class doesn't need to inherit from any other class. + +
++class PapayaWhip: ① + pass ②-
+
- The name of this class is
PapayaWhip, and it doesn't inherit from any other class. Class names are usually capitalized,EachWordLikeThis, but this is only a convention, not a requirement. + - You probably guessed this, but everything in a class is indented, just like the code within a function,
ifstatement,forloop, or any other block of code. The first line not indented is outside the class. + +This
PapayaWhipclass doesn't define any methods or attributes, but syntactically, there needs to be something in the definition, thus thepassstatement. This is a Python reserved word that just means “move along, nothing to see here”. It's a statement that does nothing, and it's a good placeholder when you're stubbing out functions or classes. + ++
+ +☞The
passstatement in Python is like a empty set of curly braces ({}) in Java or C. +Many classes are inherited from other classes, but this one is not. Many classes define methods, but this one does not. There is nothing that a Python class absolutely must have, other than a name. In particular, C++ programmers may find it odd that Python classes don't have explicit constructors and destructors. Although it's not required, Python classes can have something similar to a constructor: the
__init__method. + +The
+ +__init__()MethodFIXME - port from DiP + +
Know When To Use
+ +selfand__init__FIXME - port from DiP + +
Instantiating Classes
+ +FIXME - port from DiP + +
A Note About Garbage Collection
+ +FIXME - port from DiP, verify it's still true + +
Special Method Names
+ +FIXME - port from DiP, link to http://docs.python.org/3.0/reference/datamodel.html#special-method-names + +
FIXME - do we want to make an appendix out of some of the special methods? The organization in the Python docs is somewhat haphazard and most names have no examples at all + +
Class Attributes
+ +FIXME + +
A Fibonacci Iterator
+ +FIXME + +
+class Fib: ① + def __init__(self, max): ② + self.max = max + + def __iter__(self): ③ + self.a, self.b = 0, 1 + return self + + def __next__(self): ④ + fib = self.a + if fib > self.max: + raise StopIteration ⑤ + self.a, self.b = self.b, self.a + self.b + return fib ⑥-
+
- To build an iterator from scratch,
fibneeds to be a class, not a function. + - “Calling”
fib(max)is really creating an instance of this class and calling its__init__()method with max. The__init__()method saves the maximum value as an instance variable so other methods can refer to it later. + - The
__iter__()method is called whenever someone callsiter(fib). (As you’ll see in a minute, aforloop will call this automatically, but you can also call it yourself manually.) After performing beginning-of-iteration initialization (in this case, resettingself.aandself.b, our two counters), the__iter__()method can return any object that implements a__next__()method. In this case (and in most cases),__iter__()simply returnsself, since this class implements its own__next__()method. + - The
__next__()method is called whenever someone callsnext()on an iterator of an instance of a class. That will make more sense in a minute. + - When the
__next__()method raises aStopIterationexception, this signals to the caller that the iteration is over; no more values are available. If the caller is aforloop, it will notice thisStopIterationexception and gracefully exit the loop. (In other words, it will swallow the exception.) This little bit of magic is actually the key to using iterators inforloops. + - To spit out the next value, an iterator’s
__next__()method simplyreturns the value. Do not useyieldhere; that’s a bit of syntactic sugar that only applies when you’re using generators. Here you’re creating your own iterator from scratch; usereturninstead. +
Thoroughly confused yet? Excellent. Let’s see how to call this iterator:
+ ++>>> from fibonacci2 import Fib +>>> for n in Fib(1000): +... print(n, end=' ') +0 1 1 2 3 5 8 13 21 34 55 89 144 233 377 610 987
+ +Why, it’s exactly the same! Byte for byte identical to how you called Fibonacci-as-a-generator (modulo one capital letter). But how? + +
There’s a bit of magic involved in
forloops. Here’s what happens: + +-
+
- The
forloop callsFib(1000), as shown. This returns an instance of theFibclass. Call this fib_inst. + - Secretly, and quite cleverly, the
forloop callsiter(fib_inst), which returns an iterator object. Call this fib_iter. In this case, fib_iter == fib_inst, because the__iter__()method returnsself, but theforloop doesn’t know (or care) about that. + - To “loop through” the iterator, the
forloop callsnext(fib_iter), which calls the__next__()method on thefib_iterobject, which does the next-Fibonacci-number calculations and returns a value. Theforloop takes this value and assigns it to n, then executes the body of theforloop for that value of n. + - How does the
forloop know when to stop? I’m glad you asked! Whennext(fib_iter)raises aStopIterationexception, theforloop will swallow the exception and gracefully exit. (Any other exception will pass through and be raised as usual.) And where have you seen aStopIterationexception? In the__next__()method, of course! +
A Plural Rule Iterator
+ + +Now it’s time for the finale. Let's rewrite the plural rules generator as an iterator. + +
+ +class LazyRules: + rules_filename = 'plural6-rules.txt' + + def __init__(self): + self.pattern_file = open(self.rules_filename) + self.cache = [] + + def __iter__(self): + self.cache_index = 0 + return self + + def __next__(self): + self.cache_index += 1 + if len(self.cache) >= self.cache_index: + return self.cache[self.cache_index - 1] + + if self.pattern_file.closed: + raise StopIteration + + line = self.pattern_file.readline() + if not line: + self.pattern_file.close() + raise StopIteration + + pattern, search, replace = line.split(None, 3) + funcs = build_match_and_apply_functions( + pattern, search, replace) + self.cache.append(funcs) + return funcs + +rules = LazyRules()So this is a class that implements
__iter__()and__next__(), so it can be used as an iterator. Then, you instantiate the class and assign it to rules. This happens just once, on import. + +Let’s take the class one bite at a time. + +
+class LazyRules: + def __init__(self): ① + self.pattern_file = open('plural6-rules.txt') ③ + self.cache = [] ②-
+
- The
__init__()method is only going to be called once, when you instantiate the class and assign it to rules. + - Since this is only going to get called once, it’s the perfect place to open the pattern file. You’ll read it later; no point doing more than you absolutely have to until absolutely necessary! +
- Also, this is a good place to initialize the cache, which you’ll use later as you read the patterns from the pattern file. +
+def __iter__(self): ① + self.cache_index = 0 ② + return self ③ +-
+
- The
__iter__()method will be called every time someone — say, aforloop — callsiter(rules). + - This is the place to reset the counter that we’re going to use to retrieve items from the cache (that we haven’t built yet — patience, grasshopper). +
- Finally, the
__iter__()method returnsself, which signals that this class will take care of returning its own values throughout an iteration. +
+def __next__(self): ① + . + . + . + pattern, search, replace = line.split(None, 3) + funcs = build_match_and_apply_functions( ② + pattern, search, replace) + self.cache.append(funcs) ③ + return funcs-
+
- The
__next__()method gets called whenever someone — say, aforloop — callsnext(rules). This method will only make sense if we start at the end and work backwards. So let’s do that. + - The last part of this function should look familiar, at least. The
build_match_and_apply_functions()function hasn’t changed; it’s the same as it ever was. Each line of the pattern file will be read exactly once, as late as possible. + - The only difference is that, before returning the match and apply functions (which are stored in the tuple funcs), we’ve going to save them in
self.cache. Each match and apply function will be built exactly once, as late as possible, then cached. +
Moving backwards… + +
+def __next__(self): + . + . + . + line = self.pattern_file.readline() ① + if not line: ② + self.pattern_file.close() + raise StopIteration ③ + . + . + .-
+
- A bit of advanced file trickery here. The
readline()method (note: singular, not the pluralreadlines()) reads exactly one line from an open file. Specifically, the next line. (File objects are iterators too! It’s iterators all the way down…) + - If there was a line for
readline()to read, line will not be an empty string. Even if the file contained a blank line, line would end up as the one-character string'\n'(a carriage return). If line is really an empty string, that means there are no more lines to read from the file. + - When we reach the end of the file, we should close the file and raise the magic
StopIterationexception. Remember, we got to this point because we needed a match and apply function for the next rule. The next rule comes from the next line of the file… but there is no next line! Therefore, we have no value to return. The iteration is over. (♫ The party’s over… ♫) +
Moving backwards all the way to the start of the
__next__()method… + +
+def __next__(self): + self.cache_index += 1 + if len(self.cache) >= self.cache_index: + return self.cache[self.cache_index - 1] ① + + if self.pattern_file.closed: + raise StopIteration ② + . + . + .-
+
self.cachewill be a list of the functions we need to match and apply individual rules. (At least that should sound familiar!)self.cache_indexkeeps track of which cached item we should return next. If we haven’t exhausted the cache yet (i.e. if the length ofself.cacheis greater thanself.cache_index), then we have a cache hit! Hooray! We can return the match and apply functions from the cache instead of building them from scratch. +- On the other hand, if we don’t get a hit from the cache, and the file object has been closed (which could happen, further down the method, as you saw in the previous code snippet), then there’s nothing more we can do. If the file is closed, it means we’ve exhausted it — we’ve already read through every line from the pattern file, and we’ve already built and cached the match and apply functions for each pattern. The file is exhausted; the cache is exhausted; I’m exhausted. Wait, what? Hang in there, we’re almost done. +
Putting it all together, here’s what happens when: + +
-
+
- When the module is imported, it creates a single instance of the
LazyRulesclass, called rules, which opens the pattern file but does not read from it. + - When asked for the first match and apply function, it checks its cache but finds the cache is empty. So it reads a single line from the pattern file, builds the match and apply functions from those patterns, and caches them. +
- Let’s say, for the sake of argument, that the very first rule matched. If so, no further match and apply functions are built, and no further lines are read from the pattern file. +
- Furthermore, for the sake of argument, suppose that the caller calls the
plural()function again to pluralize a different word. Theforloop in theplural()function will calliter(rules), which will reset the cache index but will not reset the open file object. + - The first time through, the
forloop will ask for a value from rules, which will invoke its__next__()method. This time, however, the cache is primed with a single pair of match and apply functions, corresponding to the patterns in the first line of the pattern file. Since they were built and cached in the course of pluralizing the previous word, they’re retrieved from the cache. The cache index increments, and the open file is never touched. + - Let’s say, for the sake of argument, that the first rule does not match this time around. So the
forloop comes around again and asks for another value from rules. This invokes the__next__()method a second time. This time, the cache is exhausted — it only contained one item, and we’re asking for a second — so the__next__()method continues. It reads another line from the open file, builds match and apply functions out of the patterns, and caches them. + - This read-build-and-cache process will continue as long as the rules being read from the pattern file don’t match the word we’re trying to pluralize. If we do find a matching rule before the end of the file, we simply use it and stop, with the file still open. The file pointer will stay wherever we stopped reading, waiting for the next
readline()command. In the meantime, the cache now has more items in it, and if we start all over again trying to pluralize a new word, each of those items in the cache will be tried before reading the next line from the pattern file. +
Thus, we have achieved our combined goal [FIXME xref]: + +
-
+
- Minimal startup cost. The only thing that happens on
importis instantiating a single class and opening a file (but not reading from it). + - Maximum performance. The previous example would read through the file and build functions dynamically every time you wanted to pluralize a word. This version will cache functions as soon as they’re built, and in the worst case, it will only read through the pattern file once, no matter how many words you pluralize. +
- Separation of code and data. All the patterns are stored in a separate file. Code is code, and data is data, and never the twain shall meet. +
Further Reading
+ + +© 2001–9 Mark Pilgrim + + diff --git a/regular-expressions.html b/regular-expressions.html index 8b56fe0..ad041dc 100644 --- a/regular-expressions.html +++ b/regular-expressions.html @@ -417,7 +417,7 @@ body{counter-reset:h1 4}
- To build an iterator from scratch,
(x)in general is a remembered group. You can get the value of what matched by using thegroups()method of the object returned byre.search.Regular expressions are extremely powerful, but they are not the correct solution for every problem. You should learn enough about them to know when they are appropriate, when they will solve your problems, and when they will cause more problems than they solve. -
© 2001–9 Mark Pilgrim diff --git a/table-of-contents.html b/table-of-contents.html index c569516..e2af73a 100644 --- a/table-of-contents.html +++ b/table-of-contents.html @@ -119,38 +119,44 @@ ul li ol{margin:0;padding:0 0 0 2.5em}
- Case study: parsing phone numbers
- Summary
- Iterators & generators +
- Generators -
- Advanced Iterators +
- Iterators +
- Advanced Iterators +
-
+
- Diving In +
- Finding all occurrences of a pattern +
- Finding the unique items in a sequence +
- Making assertions +
- Generator expressions +
- Calculating Permutations… The Lazy Way! +
- Other Fun Stuff in the
itertoolsModule + - A New Kind Of String Manipulation +
- Evaluating Arbitrary Strings As Python Expressions +
- Putting It All Together +
- Further Reading
- Objects and object-orientation
- ...major changes afoot... -
- ...stuff about decorators... -
- ...stuff about importing modules...
-
-
-
- ...mention why from module import * is only allowed at module level -
- Unit testing
- (Not) diving in
@@ -360,11 +366,15 @@ ul li ol{margin:0;padding:0 0 0 2.5em}
Orphans (not sure where these belong yet):
- Tuples -
- Iterators -
- Generators
- List comprehensions
- Set comprehensions
- Dictionary comprehensions
- Views (several dictionary methods return them, they're dynamic, update when the dictionary changes, etc.) +
- Decorators +
- Importing modules
+
-
+
- ...mention why from module import * is only allowed at module level +
© 2001–9 Mark Pilgrim
- (Not) diving in
@@ -360,11 +366,15 @@ ul li ol{margin:0;padding:0 0 0 2.5em}
- The name of this class is