1tutorial_tests = """
2Let's try a simple generator:
3
4    >>> def f():
5    ...    yield 1
6    ...    yield 2
7
8    >>> for i in f():
9    ...     print i
10    1
11    2
12    >>> g = f()
13    >>> g.next()
14    1
15    >>> g.next()
16    2
17
18"Falling off the end" stops the generator:
19
20    >>> g.next()
21    Traceback (most recent call last):
22      File "<stdin>", line 1, in ?
23      File "<stdin>", line 2, in g
24    StopIteration
25
26"return" also stops the generator:
27
28    >>> def f():
29    ...     yield 1
30    ...     return
31    ...     yield 2 # never reached
32    ...
33    >>> g = f()
34    >>> g.next()
35    1
36    >>> g.next()
37    Traceback (most recent call last):
38      File "<stdin>", line 1, in ?
39      File "<stdin>", line 3, in f
40    StopIteration
41    >>> g.next() # once stopped, can't be resumed
42    Traceback (most recent call last):
43      File "<stdin>", line 1, in ?
44    StopIteration
45
46"raise StopIteration" stops the generator too:
47
48    >>> def f():
49    ...     yield 1
50    ...     raise StopIteration
51    ...     yield 2 # never reached
52    ...
53    >>> g = f()
54    >>> g.next()
55    1
56    >>> g.next()
57    Traceback (most recent call last):
58      File "<stdin>", line 1, in ?
59    StopIteration
60    >>> g.next()
61    Traceback (most recent call last):
62      File "<stdin>", line 1, in ?
63    StopIteration
64
65However, they are not exactly equivalent:
66
67    >>> def g1():
68    ...     try:
69    ...         return
70    ...     except:
71    ...         yield 1
72    ...
73    >>> list(g1())
74    []
75
76    >>> def g2():
77    ...     try:
78    ...         raise StopIteration
79    ...     except:
80    ...         yield 42
81    >>> print list(g2())
82    [42]
83
84This may be surprising at first:
85
86    >>> def g3():
87    ...     try:
88    ...         return
89    ...     finally:
90    ...         yield 1
91    ...
92    >>> list(g3())
93    [1]
94
95Let's create an alternate range() function implemented as a generator:
96
97    >>> def yrange(n):
98    ...     for i in range(n):
99    ...         yield i
100    ...
101    >>> list(yrange(5))
102    [0, 1, 2, 3, 4]
103
104Generators always return to the most recent caller:
105
106    >>> def creator():
107    ...     r = yrange(5)
108    ...     print "creator", r.next()
109    ...     return r
110    ...
111    >>> def caller():
112    ...     r = creator()
113    ...     for i in r:
114    ...             print "caller", i
115    ...
116    >>> caller()
117    creator 0
118    caller 1
119    caller 2
120    caller 3
121    caller 4
122
123Generators can call other generators:
124
125    >>> def zrange(n):
126    ...     for i in yrange(n):
127    ...         yield i
128    ...
129    >>> list(zrange(5))
130    [0, 1, 2, 3, 4]
131
132"""
133
134# The examples from PEP 255.
135
136pep_tests = """
137
138Specification:  Yield
139
140    Restriction:  A generator cannot be resumed while it is actively
141    running:
142
143    >>> def g():
144    ...     i = me.next()
145    ...     yield i
146    >>> me = g()
147    >>> me.next()
148    Traceback (most recent call last):
149     ...
150      File "<string>", line 2, in g
151    ValueError: generator already executing
152
153Specification: Return
154
155    Note that return isn't always equivalent to raising StopIteration:  the
156    difference lies in how enclosing try/except constructs are treated.
157    For example,
158
159        >>> def f1():
160        ...     try:
161        ...         return
162        ...     except:
163        ...        yield 1
164        >>> print list(f1())
165        []
166
167    because, as in any function, return simply exits, but
168
169        >>> def f2():
170        ...     try:
171        ...         raise StopIteration
172        ...     except:
173        ...         yield 42
174        >>> print list(f2())
175        [42]
176
177    because StopIteration is captured by a bare "except", as is any
178    exception.
179
180Specification: Generators and Exception Propagation
181
182    >>> def f():
183    ...     return 1//0
184    >>> def g():
185    ...     yield f()  # the zero division exception propagates
186    ...     yield 42   # and we'll never get here
187    >>> k = g()
188    >>> k.next()
189    Traceback (most recent call last):
190      File "<stdin>", line 1, in ?
191      File "<stdin>", line 2, in g
192      File "<stdin>", line 2, in f
193    ZeroDivisionError: integer division or modulo by zero
194    >>> k.next()  # and the generator cannot be resumed
195    Traceback (most recent call last):
196      File "<stdin>", line 1, in ?
197    StopIteration
198    >>>
199
200Specification: Try/Except/Finally
201
202    >>> def f():
203    ...     try:
204    ...         yield 1
205    ...         try:
206    ...             yield 2
207    ...             1//0
208    ...             yield 3  # never get here
209    ...         except ZeroDivisionError:
210    ...             yield 4
211    ...             yield 5
212    ...             raise
213    ...         except:
214    ...             yield 6
215    ...         yield 7     # the "raise" above stops this
216    ...     except:
217    ...         yield 8
218    ...     yield 9
219    ...     try:
220    ...         x = 12
221    ...     finally:
222    ...         yield 10
223    ...     yield 11
224    >>> print list(f())
225    [1, 2, 4, 5, 8, 9, 10, 11]
226    >>>
227
228Guido's binary tree example.
229
230    >>> # A binary tree class.
231    >>> class Tree:
232    ...
233    ...     def __init__(self, label, left=None, right=None):
234    ...         self.label = label
235    ...         self.left = left
236    ...         self.right = right
237    ...
238    ...     def __repr__(self, level=0, indent="    "):
239    ...         s = level*indent + repr(self.label)
240    ...         if self.left:
241    ...             s = s + "\\n" + self.left.__repr__(level+1, indent)
242    ...         if self.right:
243    ...             s = s + "\\n" + self.right.__repr__(level+1, indent)
244    ...         return s
245    ...
246    ...     def __iter__(self):
247    ...         return inorder(self)
248
249    >>> # Create a Tree from a list.
250    >>> def tree(list):
251    ...     n = len(list)
252    ...     if n == 0:
253    ...         return []
254    ...     i = n // 2
255    ...     return Tree(list[i], tree(list[:i]), tree(list[i+1:]))
256
257    >>> # Show it off: create a tree.
258    >>> t = tree("ABCDEFGHIJKLMNOPQRSTUVWXYZ")
259
260    >>> # A recursive generator that generates Tree labels in in-order.
261    >>> def inorder(t):
262    ...     if t:
263    ...         for x in inorder(t.left):
264    ...             yield x
265    ...         yield t.label
266    ...         for x in inorder(t.right):
267    ...             yield x
268
269    >>> # Show it off: create a tree.
270    >>> t = tree("ABCDEFGHIJKLMNOPQRSTUVWXYZ")
271    >>> # Print the nodes of the tree in in-order.
272    >>> for x in t:
273    ...     print x,
274    A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
275
276    >>> # A non-recursive generator.
277    >>> def inorder(node):
278    ...     stack = []
279    ...     while node:
280    ...         while node.left:
281    ...             stack.append(node)
282    ...             node = node.left
283    ...         yield node.label
284    ...         while not node.right:
285    ...             try:
286    ...                 node = stack.pop()
287    ...             except IndexError:
288    ...                 return
289    ...             yield node.label
290    ...         node = node.right
291
292    >>> # Exercise the non-recursive generator.
293    >>> for x in t:
294    ...     print x,
295    A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
296
297"""
298
299# Examples from Iterator-List and Python-Dev and c.l.py.
300
301email_tests = """
302
303The difference between yielding None and returning it.
304
305>>> def g():
306...     for i in range(3):
307...         yield None
308...     yield None
309...     return
310>>> list(g())
311[None, None, None, None]
312
313Ensure that explicitly raising StopIteration acts like any other exception
314in try/except, not like a return.
315
316>>> def g():
317...     yield 1
318...     try:
319...         raise StopIteration
320...     except:
321...         yield 2
322...     yield 3
323>>> list(g())
324[1, 2, 3]
325
326Next one was posted to c.l.py.
327
328>>> def gcomb(x, k):
329...     "Generate all combinations of k elements from list x."
330...
331...     if k > len(x):
332...         return
333...     if k == 0:
334...         yield []
335...     else:
336...         first, rest = x[0], x[1:]
337...         # A combination does or doesn't contain first.
338...         # If it does, the remainder is a k-1 comb of rest.
339...         for c in gcomb(rest, k-1):
340...             c.insert(0, first)
341...             yield c
342...         # If it doesn't contain first, it's a k comb of rest.
343...         for c in gcomb(rest, k):
344...             yield c
345
346>>> seq = range(1, 5)
347>>> for k in range(len(seq) + 2):
348...     print "%d-combs of %s:" % (k, seq)
349...     for c in gcomb(seq, k):
350...         print "   ", c
3510-combs of [1, 2, 3, 4]:
352    []
3531-combs of [1, 2, 3, 4]:
354    [1]
355    [2]
356    [3]
357    [4]
3582-combs of [1, 2, 3, 4]:
359    [1, 2]
360    [1, 3]
361    [1, 4]
362    [2, 3]
363    [2, 4]
364    [3, 4]
3653-combs of [1, 2, 3, 4]:
366    [1, 2, 3]
367    [1, 2, 4]
368    [1, 3, 4]
369    [2, 3, 4]
3704-combs of [1, 2, 3, 4]:
371    [1, 2, 3, 4]
3725-combs of [1, 2, 3, 4]:
373
374From the Iterators list, about the types of these things.
375
376>>> def g():
377...     yield 1
378...
379>>> type(g)
380<type 'function'>
381>>> i = g()
382>>> type(i)
383<type 'generator'>
384>>> [s for s in dir(i) if not s.startswith('_')]
385['close', 'gi_code', 'gi_frame', 'gi_running', 'next', 'send', 'throw']
386>>> from test.test_support import HAVE_DOCSTRINGS
387>>> print(i.next.__doc__ if HAVE_DOCSTRINGS else 'x.next() -> the next value, or raise StopIteration')
388x.next() -> the next value, or raise StopIteration
389>>> iter(i) is i
390True
391>>> import types
392>>> isinstance(i, types.GeneratorType)
393True
394
395And more, added later.
396
397>>> i.gi_running
3980
399>>> type(i.gi_frame)
400<type 'frame'>
401>>> i.gi_running = 42
402Traceback (most recent call last):
403  ...
404TypeError: readonly attribute
405>>> def g():
406...     yield me.gi_running
407>>> me = g()
408>>> me.gi_running
4090
410>>> me.next()
4111
412>>> me.gi_running
4130
414
415A clever union-find implementation from c.l.py, due to David Eppstein.
416Sent: Friday, June 29, 2001 12:16 PM
417To: python-list@python.org
418Subject: Re: PEP 255: Simple Generators
419
420>>> class disjointSet:
421...     def __init__(self, name):
422...         self.name = name
423...         self.parent = None
424...         self.generator = self.generate()
425...
426...     def generate(self):
427...         while not self.parent:
428...             yield self
429...         for x in self.parent.generator:
430...             yield x
431...
432...     def find(self):
433...         return self.generator.next()
434...
435...     def union(self, parent):
436...         if self.parent:
437...             raise ValueError("Sorry, I'm not a root!")
438...         self.parent = parent
439...
440...     def __str__(self):
441...         return self.name
442
443>>> names = "ABCDEFGHIJKLM"
444>>> sets = [disjointSet(name) for name in names]
445>>> roots = sets[:]
446
447>>> import random
448>>> gen = random.WichmannHill(42)
449>>> while 1:
450...     for s in sets:
451...         print "%s->%s" % (s, s.find()),
452...     print
453...     if len(roots) > 1:
454...         s1 = gen.choice(roots)
455...         roots.remove(s1)
456...         s2 = gen.choice(roots)
457...         s1.union(s2)
458...         print "merged", s1, "into", s2
459...     else:
460...         break
461A->A B->B C->C D->D E->E F->F G->G H->H I->I J->J K->K L->L M->M
462merged D into G
463A->A B->B C->C D->G E->E F->F G->G H->H I->I J->J K->K L->L M->M
464merged C into F
465A->A B->B C->F D->G E->E F->F G->G H->H I->I J->J K->K L->L M->M
466merged L into A
467A->A B->B C->F D->G E->E F->F G->G H->H I->I J->J K->K L->A M->M
468merged H into E
469A->A B->B C->F D->G E->E F->F G->G H->E I->I J->J K->K L->A M->M
470merged B into E
471A->A B->E C->F D->G E->E F->F G->G H->E I->I J->J K->K L->A M->M
472merged J into G
473A->A B->E C->F D->G E->E F->F G->G H->E I->I J->G K->K L->A M->M
474merged E into G
475A->A B->G C->F D->G E->G F->F G->G H->G I->I J->G K->K L->A M->M
476merged M into G
477A->A B->G C->F D->G E->G F->F G->G H->G I->I J->G K->K L->A M->G
478merged I into K
479A->A B->G C->F D->G E->G F->F G->G H->G I->K J->G K->K L->A M->G
480merged K into A
481A->A B->G C->F D->G E->G F->F G->G H->G I->A J->G K->A L->A M->G
482merged F into A
483A->A B->G C->A D->G E->G F->A G->G H->G I->A J->G K->A L->A M->G
484merged A into G
485A->G B->G C->G D->G E->G F->G G->G H->G I->G J->G K->G L->G M->G
486
487"""
488# Emacs turd '
489
490# Fun tests (for sufficiently warped notions of "fun").
491
492fun_tests = """
493
494Build up to a recursive Sieve of Eratosthenes generator.
495
496>>> def firstn(g, n):
497...     return [g.next() for i in range(n)]
498
499>>> def intsfrom(i):
500...     while 1:
501...         yield i
502...         i += 1
503
504>>> firstn(intsfrom(5), 7)
505[5, 6, 7, 8, 9, 10, 11]
506
507>>> def exclude_multiples(n, ints):
508...     for i in ints:
509...         if i % n:
510...             yield i
511
512>>> firstn(exclude_multiples(3, intsfrom(1)), 6)
513[1, 2, 4, 5, 7, 8]
514
515>>> def sieve(ints):
516...     prime = ints.next()
517...     yield prime
518...     not_divisible_by_prime = exclude_multiples(prime, ints)
519...     for p in sieve(not_divisible_by_prime):
520...         yield p
521
522>>> primes = sieve(intsfrom(2))
523>>> firstn(primes, 20)
524[2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71]
525
526
527Another famous problem:  generate all integers of the form
528    2**i * 3**j  * 5**k
529in increasing order, where i,j,k >= 0.  Trickier than it may look at first!
530Try writing it without generators, and correctly, and without generating
5313 internal results for each result output.
532
533>>> def times(n, g):
534...     for i in g:
535...         yield n * i
536>>> firstn(times(10, intsfrom(1)), 10)
537[10, 20, 30, 40, 50, 60, 70, 80, 90, 100]
538
539>>> def merge(g, h):
540...     ng = g.next()
541...     nh = h.next()
542...     while 1:
543...         if ng < nh:
544...             yield ng
545...             ng = g.next()
546...         elif ng > nh:
547...             yield nh
548...             nh = h.next()
549...         else:
550...             yield ng
551...             ng = g.next()
552...             nh = h.next()
553
554The following works, but is doing a whale of a lot of redundant work --
555it's not clear how to get the internal uses of m235 to share a single
556generator.  Note that me_times2 (etc) each need to see every element in the
557result sequence.  So this is an example where lazy lists are more natural
558(you can look at the head of a lazy list any number of times).
559
560>>> def m235():
561...     yield 1
562...     me_times2 = times(2, m235())
563...     me_times3 = times(3, m235())
564...     me_times5 = times(5, m235())
565...     for i in merge(merge(me_times2,
566...                          me_times3),
567...                    me_times5):
568...         yield i
569
570Don't print "too many" of these -- the implementation above is extremely
571inefficient:  each call of m235() leads to 3 recursive calls, and in
572turn each of those 3 more, and so on, and so on, until we've descended
573enough levels to satisfy the print stmts.  Very odd:  when I printed 5
574lines of results below, this managed to screw up Win98's malloc in "the
575usual" way, i.e. the heap grew over 4Mb so Win98 started fragmenting
576address space, and it *looked* like a very slow leak.
577
578>>> result = m235()
579>>> for i in range(3):
580...     print firstn(result, 15)
581[1, 2, 3, 4, 5, 6, 8, 9, 10, 12, 15, 16, 18, 20, 24]
582[25, 27, 30, 32, 36, 40, 45, 48, 50, 54, 60, 64, 72, 75, 80]
583[81, 90, 96, 100, 108, 120, 125, 128, 135, 144, 150, 160, 162, 180, 192]
584
585Heh.  Here's one way to get a shared list, complete with an excruciating
586namespace renaming trick.  The *pretty* part is that the times() and merge()
587functions can be reused as-is, because they only assume their stream
588arguments are iterable -- a LazyList is the same as a generator to times().
589
590>>> class LazyList:
591...     def __init__(self, g):
592...         self.sofar = []
593...         self.fetch = g.next
594...
595...     def __getitem__(self, i):
596...         sofar, fetch = self.sofar, self.fetch
597...         while i >= len(sofar):
598...             sofar.append(fetch())
599...         return sofar[i]
600
601>>> def m235():
602...     yield 1
603...     # Gack:  m235 below actually refers to a LazyList.
604...     me_times2 = times(2, m235)
605...     me_times3 = times(3, m235)
606...     me_times5 = times(5, m235)
607...     for i in merge(merge(me_times2,
608...                          me_times3),
609...                    me_times5):
610...         yield i
611
612Print as many of these as you like -- *this* implementation is memory-
613efficient.
614
615>>> m235 = LazyList(m235())
616>>> for i in range(5):
617...     print [m235[j] for j in range(15*i, 15*(i+1))]
618[1, 2, 3, 4, 5, 6, 8, 9, 10, 12, 15, 16, 18, 20, 24]
619[25, 27, 30, 32, 36, 40, 45, 48, 50, 54, 60, 64, 72, 75, 80]
620[81, 90, 96, 100, 108, 120, 125, 128, 135, 144, 150, 160, 162, 180, 192]
621[200, 216, 225, 240, 243, 250, 256, 270, 288, 300, 320, 324, 360, 375, 384]
622[400, 405, 432, 450, 480, 486, 500, 512, 540, 576, 600, 625, 640, 648, 675]
623
624Ye olde Fibonacci generator, LazyList style.
625
626>>> def fibgen(a, b):
627...
628...     def sum(g, h):
629...         while 1:
630...             yield g.next() + h.next()
631...
632...     def tail(g):
633...         g.next()    # throw first away
634...         for x in g:
635...             yield x
636...
637...     yield a
638...     yield b
639...     for s in sum(iter(fib),
640...                  tail(iter(fib))):
641...         yield s
642
643>>> fib = LazyList(fibgen(1, 2))
644>>> firstn(iter(fib), 17)
645[1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610, 987, 1597, 2584]
646
647
648Running after your tail with itertools.tee (new in version 2.4)
649
650The algorithms "m235" (Hamming) and Fibonacci presented above are both
651examples of a whole family of FP (functional programming) algorithms
652where a function produces and returns a list while the production algorithm
653suppose the list as already produced by recursively calling itself.
654For these algorithms to work, they must:
655
656- produce at least a first element without presupposing the existence of
657  the rest of the list
658- produce their elements in a lazy manner
659
660To work efficiently, the beginning of the list must not be recomputed over
661and over again. This is ensured in most FP languages as a built-in feature.
662In python, we have to explicitly maintain a list of already computed results
663and abandon genuine recursivity.
664
665This is what had been attempted above with the LazyList class. One problem
666with that class is that it keeps a list of all of the generated results and
667therefore continually grows. This partially defeats the goal of the generator
668concept, viz. produce the results only as needed instead of producing them
669all and thereby wasting memory.
670
671Thanks to itertools.tee, it is now clear "how to get the internal uses of
672m235 to share a single generator".
673
674>>> from itertools import tee
675>>> def m235():
676...     def _m235():
677...         yield 1
678...         for n in merge(times(2, m2),
679...                        merge(times(3, m3),
680...                              times(5, m5))):
681...             yield n
682...     m1 = _m235()
683...     m2, m3, m5, mRes = tee(m1, 4)
684...     return mRes
685
686>>> it = m235()
687>>> for i in range(5):
688...     print firstn(it, 15)
689[1, 2, 3, 4, 5, 6, 8, 9, 10, 12, 15, 16, 18, 20, 24]
690[25, 27, 30, 32, 36, 40, 45, 48, 50, 54, 60, 64, 72, 75, 80]
691[81, 90, 96, 100, 108, 120, 125, 128, 135, 144, 150, 160, 162, 180, 192]
692[200, 216, 225, 240, 243, 250, 256, 270, 288, 300, 320, 324, 360, 375, 384]
693[400, 405, 432, 450, 480, 486, 500, 512, 540, 576, 600, 625, 640, 648, 675]
694
695The "tee" function does just what we want. It internally keeps a generated
696result for as long as it has not been "consumed" from all of the duplicated
697iterators, whereupon it is deleted. You can therefore print the hamming
698sequence during hours without increasing memory usage, or very little.
699
700The beauty of it is that recursive running-after-their-tail FP algorithms
701are quite straightforwardly expressed with this Python idiom.
702
703Ye olde Fibonacci generator, tee style.
704
705>>> def fib():
706...
707...     def _isum(g, h):
708...         while 1:
709...             yield g.next() + h.next()
710...
711...     def _fib():
712...         yield 1
713...         yield 2
714...         fibTail.next() # throw first away
715...         for res in _isum(fibHead, fibTail):
716...             yield res
717...
718...     realfib = _fib()
719...     fibHead, fibTail, fibRes = tee(realfib, 3)
720...     return fibRes
721
722>>> firstn(fib(), 17)
723[1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610, 987, 1597, 2584]
724
725"""
726
727# syntax_tests mostly provokes SyntaxErrors.  Also fiddling with #if 0
728# hackery.
729
730syntax_tests = """
731
732>>> def f():
733...     return 22
734...     yield 1
735Traceback (most recent call last):
736  ..
737SyntaxError: 'return' with argument inside generator (<doctest test.test_generators.__test__.syntax[0]>, line 3)
738
739>>> def f():
740...     yield 1
741...     return 22
742Traceback (most recent call last):
743  ..
744SyntaxError: 'return' with argument inside generator (<doctest test.test_generators.__test__.syntax[1]>, line 3)
745
746"return None" is not the same as "return" in a generator:
747
748>>> def f():
749...     yield 1
750...     return None
751Traceback (most recent call last):
752  ..
753SyntaxError: 'return' with argument inside generator (<doctest test.test_generators.__test__.syntax[2]>, line 3)
754
755These are fine:
756
757>>> def f():
758...     yield 1
759...     return
760
761>>> def f():
762...     try:
763...         yield 1
764...     finally:
765...         pass
766
767>>> def f():
768...     try:
769...         try:
770...             1//0
771...         except ZeroDivisionError:
772...             yield 666
773...         except:
774...             pass
775...     finally:
776...         pass
777
778>>> def f():
779...     try:
780...         try:
781...             yield 12
782...             1//0
783...         except ZeroDivisionError:
784...             yield 666
785...         except:
786...             try:
787...                 x = 12
788...             finally:
789...                 yield 12
790...     except:
791...         return
792>>> list(f())
793[12, 666]
794
795>>> def f():
796...    yield
797>>> type(f())
798<type 'generator'>
799
800
801>>> def f():
802...    if 0:
803...        yield
804>>> type(f())
805<type 'generator'>
806
807
808>>> def f():
809...     if 0:
810...         yield 1
811>>> type(f())
812<type 'generator'>
813
814>>> def f():
815...    if "":
816...        yield None
817>>> type(f())
818<type 'generator'>
819
820>>> def f():
821...     return
822...     try:
823...         if x==4:
824...             pass
825...         elif 0:
826...             try:
827...                 1//0
828...             except SyntaxError:
829...                 pass
830...             else:
831...                 if 0:
832...                     while 12:
833...                         x += 1
834...                         yield 2 # don't blink
835...                         f(a, b, c, d, e)
836...         else:
837...             pass
838...     except:
839...         x = 1
840...     return
841>>> type(f())
842<type 'generator'>
843
844>>> def f():
845...     if 0:
846...         def g():
847...             yield 1
848...
849>>> type(f())
850<type 'NoneType'>
851
852>>> def f():
853...     if 0:
854...         class C:
855...             def __init__(self):
856...                 yield 1
857...             def f(self):
858...                 yield 2
859>>> type(f())
860<type 'NoneType'>
861
862>>> def f():
863...     if 0:
864...         return
865...     if 0:
866...         yield 2
867>>> type(f())
868<type 'generator'>
869
870
871>>> def f():
872...     if 0:
873...         lambda x:  x        # shouldn't trigger here
874...         return              # or here
875...         def f(i):
876...             return 2*i      # or here
877...         if 0:
878...             return 3        # but *this* sucks (line 8)
879...     if 0:
880...         yield 2             # because it's a generator (line 10)
881Traceback (most recent call last):
882SyntaxError: 'return' with argument inside generator (<doctest test.test_generators.__test__.syntax[24]>, line 10)
883
884This one caused a crash (see SF bug 567538):
885
886>>> def f():
887...     for i in range(3):
888...         try:
889...             continue
890...         finally:
891...             yield i
892...
893>>> g = f()
894>>> print g.next()
8950
896>>> print g.next()
8971
898>>> print g.next()
8992
900>>> print g.next()
901Traceback (most recent call last):
902StopIteration
903
904
905Test the gi_code attribute
906
907>>> def f():
908...     yield 5
909...
910>>> g = f()
911>>> g.gi_code is f.func_code
912True
913>>> g.next()
9145
915>>> g.next()
916Traceback (most recent call last):
917StopIteration
918>>> g.gi_code is f.func_code
919True
920
921
922Test the __name__ attribute and the repr()
923
924>>> def f():
925...    yield 5
926...
927>>> g = f()
928>>> g.__name__
929'f'
930>>> repr(g)  # doctest: +ELLIPSIS
931'<generator object f at ...>'
932
933Lambdas shouldn't have their usual return behavior.
934
935>>> x = lambda: (yield 1)
936>>> list(x())
937[1]
938
939>>> x = lambda: ((yield 1), (yield 2))
940>>> list(x())
941[1, 2]
942"""
943
944# conjoin is a simple backtracking generator, named in honor of Icon's
945# "conjunction" control structure.  Pass a list of no-argument functions
946# that return iterable objects.  Easiest to explain by example:  assume the
947# function list [x, y, z] is passed.  Then conjoin acts like:
948#
949# def g():
950#     values = [None] * 3
951#     for values[0] in x():
952#         for values[1] in y():
953#             for values[2] in z():
954#                 yield values
955#
956# So some 3-lists of values *may* be generated, each time we successfully
957# get into the innermost loop.  If an iterator fails (is exhausted) before
958# then, it "backtracks" to get the next value from the nearest enclosing
959# iterator (the one "to the left"), and starts all over again at the next
960# slot (pumps a fresh iterator).  Of course this is most useful when the
961# iterators have side-effects, so that which values *can* be generated at
962# each slot depend on the values iterated at previous slots.
963
964def simple_conjoin(gs):
965
966    values = [None] * len(gs)
967
968    def gen(i):
969        if i >= len(gs):
970            yield values
971        else:
972            for values[i] in gs[i]():
973                for x in gen(i+1):
974                    yield x
975
976    for x in gen(0):
977        yield x
978
979# That works fine, but recursing a level and checking i against len(gs) for
980# each item produced is inefficient.  By doing manual loop unrolling across
981# generator boundaries, it's possible to eliminate most of that overhead.
982# This isn't worth the bother *in general* for generators, but conjoin() is
983# a core building block for some CPU-intensive generator applications.
984
985def conjoin(gs):
986
987    n = len(gs)
988    values = [None] * n
989
990    # Do one loop nest at time recursively, until the # of loop nests
991    # remaining is divisible by 3.
992
993    def gen(i):
994        if i >= n:
995            yield values
996
997        elif (n-i) % 3:
998            ip1 = i+1
999            for values[i] in gs[i]():
1000                for x in gen(ip1):
1001                    yield x
1002
1003        else:
1004            for x in _gen3(i):
1005                yield x
1006
1007    # Do three loop nests at a time, recursing only if at least three more
1008    # remain.  Don't call directly:  this is an internal optimization for
1009    # gen's use.
1010
1011    def _gen3(i):
1012        assert i < n and (n-i) % 3 == 0
1013        ip1, ip2, ip3 = i+1, i+2, i+3
1014        g, g1, g2 = gs[i : ip3]
1015
1016        if ip3 >= n:
1017            # These are the last three, so we can yield values directly.
1018            for values[i] in g():
1019                for values[ip1] in g1():
1020                    for values[ip2] in g2():
1021                        yield values
1022
1023        else:
1024            # At least 6 loop nests remain; peel off 3 and recurse for the
1025            # rest.
1026            for values[i] in g():
1027                for values[ip1] in g1():
1028                    for values[ip2] in g2():
1029                        for x in _gen3(ip3):
1030                            yield x
1031
1032    for x in gen(0):
1033        yield x
1034
1035# And one more approach:  For backtracking apps like the Knight's Tour
1036# solver below, the number of backtracking levels can be enormous (one
1037# level per square, for the Knight's Tour, so that e.g. a 100x100 board
1038# needs 10,000 levels).  In such cases Python is likely to run out of
1039# stack space due to recursion.  So here's a recursion-free version of
1040# conjoin too.
1041# NOTE WELL:  This allows large problems to be solved with only trivial
1042# demands on stack space.  Without explicitly resumable generators, this is
1043# much harder to achieve.  OTOH, this is much slower (up to a factor of 2)
1044# than the fancy unrolled recursive conjoin.
1045
1046def flat_conjoin(gs):  # rename to conjoin to run tests with this instead
1047    n = len(gs)
1048    values = [None] * n
1049    iters  = [None] * n
1050    _StopIteration = StopIteration  # make local because caught a *lot*
1051    i = 0
1052    while 1:
1053        # Descend.
1054        try:
1055            while i < n:
1056                it = iters[i] = gs[i]().next
1057                values[i] = it()
1058                i += 1
1059        except _StopIteration:
1060            pass
1061        else:
1062            assert i == n
1063            yield values
1064
1065        # Backtrack until an older iterator can be resumed.
1066        i -= 1
1067        while i >= 0:
1068            try:
1069                values[i] = iters[i]()
1070                # Success!  Start fresh at next level.
1071                i += 1
1072                break
1073            except _StopIteration:
1074                # Continue backtracking.
1075                i -= 1
1076        else:
1077            assert i < 0
1078            break
1079
1080# A conjoin-based N-Queens solver.
1081
1082class Queens:
1083    def __init__(self, n):
1084        self.n = n
1085        rangen = range(n)
1086
1087        # Assign a unique int to each column and diagonal.
1088        # columns:  n of those, range(n).
1089        # NW-SE diagonals: 2n-1 of these, i-j unique and invariant along
1090        # each, smallest i-j is 0-(n-1) = 1-n, so add n-1 to shift to 0-
1091        # based.
1092        # NE-SW diagonals: 2n-1 of these, i+j unique and invariant along
1093        # each, smallest i+j is 0, largest is 2n-2.
1094
1095        # For each square, compute a bit vector of the columns and
1096        # diagonals it covers, and for each row compute a function that
1097        # generates the possibilities for the columns in that row.
1098        self.rowgenerators = []
1099        for i in rangen:
1100            rowuses = [(1L << j) |                  # column ordinal
1101                       (1L << (n + i-j + n-1)) |    # NW-SE ordinal
1102                       (1L << (n + 2*n-1 + i+j))    # NE-SW ordinal
1103                            for j in rangen]
1104
1105            def rowgen(rowuses=rowuses):
1106                for j in rangen:
1107                    uses = rowuses[j]
1108                    if uses & self.used == 0:
1109                        self.used |= uses
1110                        yield j
1111                        self.used &= ~uses
1112
1113            self.rowgenerators.append(rowgen)
1114
1115    # Generate solutions.
1116    def solve(self):
1117        self.used = 0
1118        for row2col in conjoin(self.rowgenerators):
1119            yield row2col
1120
1121    def printsolution(self, row2col):
1122        n = self.n
1123        assert n == len(row2col)
1124        sep = "+" + "-+" * n
1125        print sep
1126        for i in range(n):
1127            squares = [" " for j in range(n)]
1128            squares[row2col[i]] = "Q"
1129            print "|" + "|".join(squares) + "|"
1130            print sep
1131
1132# A conjoin-based Knight's Tour solver.  This is pretty sophisticated
1133# (e.g., when used with flat_conjoin above, and passing hard=1 to the
1134# constructor, a 200x200 Knight's Tour was found quickly -- note that we're
1135# creating 10s of thousands of generators then!), and is lengthy.
1136
1137class Knights:
1138    def __init__(self, m, n, hard=0):
1139        self.m, self.n = m, n
1140
1141        # solve() will set up succs[i] to be a list of square #i's
1142        # successors.
1143        succs = self.succs = []
1144
1145        # Remove i0 from each of its successor's successor lists, i.e.
1146        # successors can't go back to i0 again.  Return 0 if we can
1147        # detect this makes a solution impossible, else return 1.
1148
1149        def remove_from_successors(i0, len=len):
1150            # If we remove all exits from a free square, we're dead:
1151            # even if we move to it next, we can't leave it again.
1152            # If we create a square with one exit, we must visit it next;
1153            # else somebody else will have to visit it, and since there's
1154            # only one adjacent, there won't be a way to leave it again.
1155            # Finelly, if we create more than one free square with a
1156            # single exit, we can only move to one of them next, leaving
1157            # the other one a dead end.
1158            ne0 = ne1 = 0
1159            for i in succs[i0]:
1160                s = succs[i]
1161                s.remove(i0)
1162                e = len(s)
1163                if e == 0:
1164                    ne0 += 1
1165                elif e == 1:
1166                    ne1 += 1
1167            return ne0 == 0 and ne1 < 2
1168
1169        # Put i0 back in each of its successor's successor lists.
1170
1171        def add_to_successors(i0):
1172            for i in succs[i0]:
1173                succs[i].append(i0)
1174
1175        # Generate the first move.
1176        def first():
1177            if m < 1 or n < 1:
1178                return
1179
1180            # Since we're looking for a cycle, it doesn't matter where we
1181            # start.  Starting in a corner makes the 2nd move easy.
1182            corner = self.coords2index(0, 0)
1183            remove_from_successors(corner)
1184            self.lastij = corner
1185            yield corner
1186            add_to_successors(corner)
1187
1188        # Generate the second moves.
1189        def second():
1190            corner = self.coords2index(0, 0)
1191            assert self.lastij == corner  # i.e., we started in the corner
1192            if m < 3 or n < 3:
1193                return
1194            assert len(succs[corner]) == 2
1195            assert self.coords2index(1, 2) in succs[corner]
1196            assert self.coords2index(2, 1) in succs[corner]
1197            # Only two choices.  Whichever we pick, the other must be the
1198            # square picked on move m*n, as it's the only way to get back
1199            # to (0, 0).  Save its index in self.final so that moves before
1200            # the last know it must be kept free.
1201            for i, j in (1, 2), (2, 1):
1202                this  = self.coords2index(i, j)
1203                final = self.coords2index(3-i, 3-j)
1204                self.final = final
1205
1206                remove_from_successors(this)
1207                succs[final].append(corner)
1208                self.lastij = this
1209                yield this
1210                succs[final].remove(corner)
1211                add_to_successors(this)
1212
1213        # Generate moves 3 thru m*n-1.
1214        def advance(len=len):
1215            # If some successor has only one exit, must take it.
1216            # Else favor successors with fewer exits.
1217            candidates = []
1218            for i in succs[self.lastij]:
1219                e = len(succs[i])
1220                assert e > 0, "else remove_from_successors() pruning flawed"
1221                if e == 1:
1222                    candidates = [(e, i)]
1223                    break
1224                candidates.append((e, i))
1225            else:
1226                candidates.sort()
1227
1228            for e, i in candidates:
1229                if i != self.final:
1230                    if remove_from_successors(i):
1231                        self.lastij = i
1232                        yield i
1233                    add_to_successors(i)
1234
1235        # Generate moves 3 thru m*n-1.  Alternative version using a
1236        # stronger (but more expensive) heuristic to order successors.
1237        # Since the # of backtracking levels is m*n, a poor move early on
1238        # can take eons to undo.  Smallest square board for which this
1239        # matters a lot is 52x52.
1240        def advance_hard(vmid=(m-1)/2.0, hmid=(n-1)/2.0, len=len):
1241            # If some successor has only one exit, must take it.
1242            # Else favor successors with fewer exits.
1243            # Break ties via max distance from board centerpoint (favor
1244            # corners and edges whenever possible).
1245            candidates = []
1246            for i in succs[self.lastij]:
1247                e = len(succs[i])
1248                assert e > 0, "else remove_from_successors() pruning flawed"
1249                if e == 1:
1250                    candidates = [(e, 0, i)]
1251                    break
1252                i1, j1 = self.index2coords(i)
1253                d = (i1 - vmid)**2 + (j1 - hmid)**2
1254                candidates.append((e, -d, i))
1255            else:
1256                candidates.sort()
1257
1258            for e, d, i in candidates:
1259                if i != self.final:
1260                    if remove_from_successors(i):
1261                        self.lastij = i
1262                        yield i
1263                    add_to_successors(i)
1264
1265        # Generate the last move.
1266        def last():
1267            assert self.final in succs[self.lastij]
1268            yield self.final
1269
1270        if m*n < 4:
1271            self.squaregenerators = [first]
1272        else:
1273            self.squaregenerators = [first, second] + \
1274                [hard and advance_hard or advance] * (m*n - 3) + \
1275                [last]
1276
1277    def coords2index(self, i, j):
1278        assert 0 <= i < self.m
1279        assert 0 <= j < self.n
1280        return i * self.n + j
1281
1282    def index2coords(self, index):
1283        assert 0 <= index < self.m * self.n
1284        return divmod(index, self.n)
1285
1286    def _init_board(self):
1287        succs = self.succs
1288        del succs[:]
1289        m, n = self.m, self.n
1290        c2i = self.coords2index
1291
1292        offsets = [( 1,  2), ( 2,  1), ( 2, -1), ( 1, -2),
1293                   (-1, -2), (-2, -1), (-2,  1), (-1,  2)]
1294        rangen = range(n)
1295        for i in range(m):
1296            for j in rangen:
1297                s = [c2i(i+io, j+jo) for io, jo in offsets
1298                                     if 0 <= i+io < m and
1299                                        0 <= j+jo < n]
1300                succs.append(s)
1301
1302    # Generate solutions.
1303    def solve(self):
1304        self._init_board()
1305        for x in conjoin(self.squaregenerators):
1306            yield x
1307
1308    def printsolution(self, x):
1309        m, n = self.m, self.n
1310        assert len(x) == m*n
1311        w = len(str(m*n))
1312        format = "%" + str(w) + "d"
1313
1314        squares = [[None] * n for i in range(m)]
1315        k = 1
1316        for i in x:
1317            i1, j1 = self.index2coords(i)
1318            squares[i1][j1] = format % k
1319            k += 1
1320
1321        sep = "+" + ("-" * w + "+") * n
1322        print sep
1323        for i in range(m):
1324            row = squares[i]
1325            print "|" + "|".join(row) + "|"
1326            print sep
1327
1328conjoin_tests = """
1329
1330Generate the 3-bit binary numbers in order.  This illustrates dumbest-
1331possible use of conjoin, just to generate the full cross-product.
1332
1333>>> for c in conjoin([lambda: iter((0, 1))] * 3):
1334...     print c
1335[0, 0, 0]
1336[0, 0, 1]
1337[0, 1, 0]
1338[0, 1, 1]
1339[1, 0, 0]
1340[1, 0, 1]
1341[1, 1, 0]
1342[1, 1, 1]
1343
1344For efficiency in typical backtracking apps, conjoin() yields the same list
1345object each time.  So if you want to save away a full account of its
1346generated sequence, you need to copy its results.
1347
1348>>> def gencopy(iterator):
1349...     for x in iterator:
1350...         yield x[:]
1351
1352>>> for n in range(10):
1353...     all = list(gencopy(conjoin([lambda: iter((0, 1))] * n)))
1354...     print n, len(all), all[0] == [0] * n, all[-1] == [1] * n
13550 1 True True
13561 2 True True
13572 4 True True
13583 8 True True
13594 16 True True
13605 32 True True
13616 64 True True
13627 128 True True
13638 256 True True
13649 512 True True
1365
1366And run an 8-queens solver.
1367
1368>>> q = Queens(8)
1369>>> LIMIT = 2
1370>>> count = 0
1371>>> for row2col in q.solve():
1372...     count += 1
1373...     if count <= LIMIT:
1374...         print "Solution", count
1375...         q.printsolution(row2col)
1376Solution 1
1377+-+-+-+-+-+-+-+-+
1378|Q| | | | | | | |
1379+-+-+-+-+-+-+-+-+
1380| | | | |Q| | | |
1381+-+-+-+-+-+-+-+-+
1382| | | | | | | |Q|
1383+-+-+-+-+-+-+-+-+
1384| | | | | |Q| | |
1385+-+-+-+-+-+-+-+-+
1386| | |Q| | | | | |
1387+-+-+-+-+-+-+-+-+
1388| | | | | | |Q| |
1389+-+-+-+-+-+-+-+-+
1390| |Q| | | | | | |
1391+-+-+-+-+-+-+-+-+
1392| | | |Q| | | | |
1393+-+-+-+-+-+-+-+-+
1394Solution 2
1395+-+-+-+-+-+-+-+-+
1396|Q| | | | | | | |
1397+-+-+-+-+-+-+-+-+
1398| | | | | |Q| | |
1399+-+-+-+-+-+-+-+-+
1400| | | | | | | |Q|
1401+-+-+-+-+-+-+-+-+
1402| | |Q| | | | | |
1403+-+-+-+-+-+-+-+-+
1404| | | | | | |Q| |
1405+-+-+-+-+-+-+-+-+
1406| | | |Q| | | | |
1407+-+-+-+-+-+-+-+-+
1408| |Q| | | | | | |
1409+-+-+-+-+-+-+-+-+
1410| | | | |Q| | | |
1411+-+-+-+-+-+-+-+-+
1412
1413>>> print count, "solutions in all."
141492 solutions in all.
1415
1416And run a Knight's Tour on a 10x10 board.  Note that there are about
141720,000 solutions even on a 6x6 board, so don't dare run this to exhaustion.
1418
1419>>> k = Knights(10, 10)
1420>>> LIMIT = 2
1421>>> count = 0
1422>>> for x in k.solve():
1423...     count += 1
1424...     if count <= LIMIT:
1425...         print "Solution", count
1426...         k.printsolution(x)
1427...     else:
1428...         break
1429Solution 1
1430+---+---+---+---+---+---+---+---+---+---+
1431|  1| 58| 27| 34|  3| 40| 29| 10|  5|  8|
1432+---+---+---+---+---+---+---+---+---+---+
1433| 26| 35|  2| 57| 28| 33|  4|  7| 30| 11|
1434+---+---+---+---+---+---+---+---+---+---+
1435| 59|100| 73| 36| 41| 56| 39| 32|  9|  6|
1436+---+---+---+---+---+---+---+---+---+---+
1437| 74| 25| 60| 55| 72| 37| 42| 49| 12| 31|
1438+---+---+---+---+---+---+---+---+---+---+
1439| 61| 86| 99| 76| 63| 52| 47| 38| 43| 50|
1440+---+---+---+---+---+---+---+---+---+---+
1441| 24| 75| 62| 85| 54| 71| 64| 51| 48| 13|
1442+---+---+---+---+---+---+---+---+---+---+
1443| 87| 98| 91| 80| 77| 84| 53| 46| 65| 44|
1444+---+---+---+---+---+---+---+---+---+---+
1445| 90| 23| 88| 95| 70| 79| 68| 83| 14| 17|
1446+---+---+---+---+---+---+---+---+---+---+
1447| 97| 92| 21| 78| 81| 94| 19| 16| 45| 66|
1448+---+---+---+---+---+---+---+---+---+---+
1449| 22| 89| 96| 93| 20| 69| 82| 67| 18| 15|
1450+---+---+---+---+---+---+---+---+---+---+
1451Solution 2
1452+---+---+---+---+---+---+---+---+---+---+
1453|  1| 58| 27| 34|  3| 40| 29| 10|  5|  8|
1454+---+---+---+---+---+---+---+---+---+---+
1455| 26| 35|  2| 57| 28| 33|  4|  7| 30| 11|
1456+---+---+---+---+---+---+---+---+---+---+
1457| 59|100| 73| 36| 41| 56| 39| 32|  9|  6|
1458+---+---+---+---+---+---+---+---+---+---+
1459| 74| 25| 60| 55| 72| 37| 42| 49| 12| 31|
1460+---+---+---+---+---+---+---+---+---+---+
1461| 61| 86| 99| 76| 63| 52| 47| 38| 43| 50|
1462+---+---+---+---+---+---+---+---+---+---+
1463| 24| 75| 62| 85| 54| 71| 64| 51| 48| 13|
1464+---+---+---+---+---+---+---+---+---+---+
1465| 87| 98| 89| 80| 77| 84| 53| 46| 65| 44|
1466+---+---+---+---+---+---+---+---+---+---+
1467| 90| 23| 92| 95| 70| 79| 68| 83| 14| 17|
1468+---+---+---+---+---+---+---+---+---+---+
1469| 97| 88| 21| 78| 81| 94| 19| 16| 45| 66|
1470+---+---+---+---+---+---+---+---+---+---+
1471| 22| 91| 96| 93| 20| 69| 82| 67| 18| 15|
1472+---+---+---+---+---+---+---+---+---+---+
1473"""
1474
1475weakref_tests = """\
1476Generators are weakly referencable:
1477
1478>>> import weakref
1479>>> def gen():
1480...     yield 'foo!'
1481...
1482>>> wr = weakref.ref(gen)
1483>>> wr() is gen
1484True
1485>>> p = weakref.proxy(gen)
1486
1487Generator-iterators are weakly referencable as well:
1488
1489>>> gi = gen()
1490>>> wr = weakref.ref(gi)
1491>>> wr() is gi
1492True
1493>>> p = weakref.proxy(gi)
1494>>> list(p)
1495['foo!']
1496
1497"""
1498
1499coroutine_tests = """\
1500Sending a value into a started generator:
1501
1502>>> def f():
1503...     print (yield 1)
1504...     yield 2
1505>>> g = f()
1506>>> g.next()
15071
1508>>> g.send(42)
150942
15102
1511
1512Sending a value into a new generator produces a TypeError:
1513
1514>>> f().send("foo")
1515Traceback (most recent call last):
1516...
1517TypeError: can't send non-None value to a just-started generator
1518
1519
1520Yield by itself yields None:
1521
1522>>> def f(): yield
1523>>> list(f())
1524[None]
1525
1526
1527
1528An obscene abuse of a yield expression within a generator expression:
1529
1530>>> list((yield 21) for i in range(4))
1531[21, None, 21, None, 21, None, 21, None]
1532
1533And a more sane, but still weird usage:
1534
1535>>> def f(): list(i for i in [(yield 26)])
1536>>> type(f())
1537<type 'generator'>
1538
1539
1540A yield expression with augmented assignment.
1541
1542>>> def coroutine(seq):
1543...     count = 0
1544...     while count < 200:
1545...         count += yield
1546...         seq.append(count)
1547>>> seq = []
1548>>> c = coroutine(seq)
1549>>> c.next()
1550>>> print seq
1551[]
1552>>> c.send(10)
1553>>> print seq
1554[10]
1555>>> c.send(10)
1556>>> print seq
1557[10, 20]
1558>>> c.send(10)
1559>>> print seq
1560[10, 20, 30]
1561
1562
1563Check some syntax errors for yield expressions:
1564
1565>>> f=lambda: (yield 1),(yield 2)
1566Traceback (most recent call last):
1567  ...
1568  File "<doctest test.test_generators.__test__.coroutine[21]>", line 1
1569SyntaxError: 'yield' outside function
1570
1571>>> def f(): return lambda x=(yield): 1
1572Traceback (most recent call last):
1573  ...
1574SyntaxError: 'return' with argument inside generator (<doctest test.test_generators.__test__.coroutine[22]>, line 1)
1575
1576>>> def f(): x = yield = y
1577Traceback (most recent call last):
1578  ...
1579  File "<doctest test.test_generators.__test__.coroutine[23]>", line 1
1580SyntaxError: assignment to yield expression not possible
1581
1582>>> def f(): (yield bar) = y
1583Traceback (most recent call last):
1584  ...
1585  File "<doctest test.test_generators.__test__.coroutine[24]>", line 1
1586SyntaxError: can't assign to yield expression
1587
1588>>> def f(): (yield bar) += y
1589Traceback (most recent call last):
1590  ...
1591  File "<doctest test.test_generators.__test__.coroutine[25]>", line 1
1592SyntaxError: can't assign to yield expression
1593
1594
1595Now check some throw() conditions:
1596
1597>>> def f():
1598...     while True:
1599...         try:
1600...             print (yield)
1601...         except ValueError,v:
1602...             print "caught ValueError (%s)" % (v),
1603>>> import sys
1604>>> g = f()
1605>>> g.next()
1606
1607>>> g.throw(ValueError) # type only
1608caught ValueError ()
1609
1610>>> g.throw(ValueError("xyz"))  # value only
1611caught ValueError (xyz)
1612
1613>>> g.throw(ValueError, ValueError(1))   # value+matching type
1614caught ValueError (1)
1615
1616>>> g.throw(ValueError, TypeError(1))  # mismatched type, rewrapped
1617caught ValueError (1)
1618
1619>>> g.throw(ValueError, ValueError(1), None)   # explicit None traceback
1620caught ValueError (1)
1621
1622>>> g.throw(ValueError(1), "foo")       # bad args
1623Traceback (most recent call last):
1624  ...
1625TypeError: instance exception may not have a separate value
1626
1627>>> g.throw(ValueError, "foo", 23)      # bad args
1628Traceback (most recent call last):
1629  ...
1630TypeError: throw() third argument must be a traceback object
1631
1632>>> def throw(g,exc):
1633...     try:
1634...         raise exc
1635...     except:
1636...         g.throw(*sys.exc_info())
1637>>> throw(g,ValueError) # do it with traceback included
1638caught ValueError ()
1639
1640>>> g.send(1)
16411
1642
1643>>> throw(g,TypeError)  # terminate the generator
1644Traceback (most recent call last):
1645  ...
1646TypeError
1647
1648>>> print g.gi_frame
1649None
1650
1651>>> g.send(2)
1652Traceback (most recent call last):
1653  ...
1654StopIteration
1655
1656>>> g.throw(ValueError,6)       # throw on closed generator
1657Traceback (most recent call last):
1658  ...
1659ValueError: 6
1660
1661>>> f().throw(ValueError,7)     # throw on just-opened generator
1662Traceback (most recent call last):
1663  ...
1664ValueError: 7
1665
1666>>> f().throw("abc")     # throw on just-opened generator
1667Traceback (most recent call last):
1668  ...
1669TypeError: exceptions must be classes, or instances, not str
1670
1671Now let's try closing a generator:
1672
1673>>> def f():
1674...     try: yield
1675...     except GeneratorExit:
1676...         print "exiting"
1677
1678>>> g = f()
1679>>> g.next()
1680>>> g.close()
1681exiting
1682>>> g.close()  # should be no-op now
1683
1684>>> f().close()  # close on just-opened generator should be fine
1685
1686>>> def f(): yield      # an even simpler generator
1687>>> f().close()         # close before opening
1688>>> g = f()
1689>>> g.next()
1690>>> g.close()           # close normally
1691
1692And finalization:
1693
1694>>> def f():
1695...     try: yield
1696...     finally:
1697...         print "exiting"
1698
1699>>> g = f()
1700>>> g.next()
1701>>> del g
1702exiting
1703
1704>>> class context(object):
1705...    def __enter__(self): pass
1706...    def __exit__(self, *args): print 'exiting'
1707>>> def f():
1708...     with context():
1709...          yield
1710>>> g = f()
1711>>> g.next()
1712>>> del g
1713exiting
1714
1715
1716GeneratorExit is not caught by except Exception:
1717
1718>>> def f():
1719...     try: yield
1720...     except Exception: print 'except'
1721...     finally: print 'finally'
1722
1723>>> g = f()
1724>>> g.next()
1725>>> del g
1726finally
1727
1728
1729Now let's try some ill-behaved generators:
1730
1731>>> def f():
1732...     try: yield
1733...     except GeneratorExit:
1734...         yield "foo!"
1735>>> g = f()
1736>>> g.next()
1737>>> g.close()
1738Traceback (most recent call last):
1739  ...
1740RuntimeError: generator ignored GeneratorExit
1741>>> g.close()
1742
1743
1744Our ill-behaved code should be invoked during GC:
1745
1746>>> import sys, StringIO
1747>>> old, sys.stderr = sys.stderr, StringIO.StringIO()
1748>>> g = f()
1749>>> g.next()
1750>>> del g
1751>>> sys.stderr.getvalue().startswith(
1752...     "Exception RuntimeError: 'generator ignored GeneratorExit' in "
1753... )
1754True
1755>>> sys.stderr = old
1756
1757
1758And errors thrown during closing should propagate:
1759
1760>>> def f():
1761...     try: yield
1762...     except GeneratorExit:
1763...         raise TypeError("fie!")
1764>>> g = f()
1765>>> g.next()
1766>>> g.close()
1767Traceback (most recent call last):
1768  ...
1769TypeError: fie!
1770
1771
1772Ensure that various yield expression constructs make their
1773enclosing function a generator:
1774
1775>>> def f(): x += yield
1776>>> type(f())
1777<type 'generator'>
1778
1779>>> def f(): x = yield
1780>>> type(f())
1781<type 'generator'>
1782
1783>>> def f(): lambda x=(yield): 1
1784>>> type(f())
1785<type 'generator'>
1786
1787>>> def f(): x=(i for i in (yield) if (yield))
1788>>> type(f())
1789<type 'generator'>
1790
1791>>> def f(d): d[(yield "a")] = d[(yield "b")] = 27
1792>>> data = [1,2]
1793>>> g = f(data)
1794>>> type(g)
1795<type 'generator'>
1796>>> g.send(None)
1797'a'
1798>>> data
1799[1, 2]
1800>>> g.send(0)
1801'b'
1802>>> data
1803[27, 2]
1804>>> try: g.send(1)
1805... except StopIteration: pass
1806>>> data
1807[27, 27]
1808
1809"""
1810
1811refleaks_tests = """
1812Prior to adding cycle-GC support to itertools.tee, this code would leak
1813references. We add it to the standard suite so the routine refleak-tests
1814would trigger if it starts being uncleanable again.
1815
1816>>> import itertools
1817>>> def leak():
1818...     class gen:
1819...         def __iter__(self):
1820...             return self
1821...         def next(self):
1822...             return self.item
1823...     g = gen()
1824...     head, tail = itertools.tee(g)
1825...     g.item = head
1826...     return head
1827>>> it = leak()
1828
1829Make sure to also test the involvement of the tee-internal teedataobject,
1830which stores returned items.
1831
1832>>> item = it.next()
1833
1834
1835
1836This test leaked at one point due to generator finalization/destruction.
1837It was copied from Lib/test/leakers/test_generator_cycle.py before the file
1838was removed.
1839
1840>>> def leak():
1841...    def gen():
1842...        while True:
1843...            yield g
1844...    g = gen()
1845
1846>>> leak()
1847
1848
1849
1850This test isn't really generator related, but rather exception-in-cleanup
1851related. The coroutine tests (above) just happen to cause an exception in
1852the generator's __del__ (tp_del) method. We can also test for this
1853explicitly, without generators. We do have to redirect stderr to avoid
1854printing warnings and to doublecheck that we actually tested what we wanted
1855to test.
1856
1857>>> import sys, StringIO
1858>>> old = sys.stderr
1859>>> try:
1860...     sys.stderr = StringIO.StringIO()
1861...     class Leaker:
1862...         def __del__(self):
1863...             raise RuntimeError
1864...
1865...     l = Leaker()
1866...     del l
1867...     err = sys.stderr.getvalue().strip()
1868...     err.startswith(
1869...         "Exception RuntimeError: RuntimeError() in <"
1870...     )
1871...     err.endswith("> ignored")
1872...     len(err.splitlines())
1873... finally:
1874...     sys.stderr = old
1875True
1876True
18771
1878
1879
1880
1881These refleak tests should perhaps be in a testfile of their own,
1882test_generators just happened to be the test that drew these out.
1883
1884"""
1885
1886__test__ = {"tut":      tutorial_tests,
1887            "pep":      pep_tests,
1888            "email":    email_tests,
1889            "fun":      fun_tests,
1890            "syntax":   syntax_tests,
1891            "conjoin":  conjoin_tests,
1892            "weakref":  weakref_tests,
1893            "coroutine":  coroutine_tests,
1894            "refleaks": refleaks_tests,
1895            }
1896
1897# Magic test name that regrtest.py invokes *after* importing this module.
1898# This worms around a bootstrap problem.
1899# Note that doctest and regrtest both look in sys.argv for a "-v" argument,
1900# so this works as expected in both ways of running regrtest.
1901def test_main(verbose=None):
1902    from test import test_support, test_generators
1903    test_support.run_doctest(test_generators, verbose)
1904
1905# This part isn't needed for regrtest, but for running the test directly.
1906if __name__ == "__main__":
1907    test_main(1)
1908