249

Is there a simple way of testing if the generator has no items, like peek, hasNext, isEmpty, something along those lines?

7
  • Correct me if I'm wrong, but if you could make a truly generic solution to any generator, it would be the equivalent of setting breakpoints on the yield statements and having the ability to "step backward". Would that mean cloning the stack frame on yields and restoring them on StopIteration? Commented Mar 19, 2009 at 13:07
  • Well, I guess restore them StopIteration or not, but at least StopIteration would tell you it was empty. Yeah I need sleep... Commented Mar 19, 2009 at 13:19
  • 5
    I think I know why he wants this. If you're doing web development with templates, and passing the return value into a template like Cheetah or something, empty list [] is conveniently Falsey so you can do an if check on it and do special behavior for something or nothing. Generators are true even if they yield no elements. Commented Mar 8, 2011 at 5:02
  • 1
    Here's my use case... I'm using glob.iglob("filepattern") on a user-supplied wildcard pattern, and I want to warn the user if the pattern does not match any files. Sure I can work around this in various ways, but it's useful to be able to cleanly test whether the iterator came up empty or not. Commented Jan 24, 2013 at 16:48
  • May be use this solution: stackoverflow.com/a/11467686/463758 Commented Nov 17, 2015 at 20:15

26 Answers 26

154

Suggestion:

def peek(iterable): try: first = next(iterable) except StopIteration: return None return first, itertools.chain([first], iterable) 

Usage:

res = peek(mysequence) if res is None: # sequence is empty. Do stuff. else: first, mysequence = res # Do something with first, maybe? # Then iterate over the sequence: for element in mysequence: # etc. 
Sign up to request clarification or add additional context in comments.

8 Comments

I don't quite get the point of returning the first element twice in return first, itertools.chain([first], rest).
@njzk2 I was going for a "peek" operation (hence the function name). wiki "peek is an operation which returns the value of the top of the collection without removing the value from the data"
This won't work if the generator is designed to yield None. def gen(): for pony in range(4): yield None if pony == 2 else pony
@Paul Look at the return values closely. If the generator is done -- i.e., not returning None, but raising StopIteration -- the result of the function is None. Otherwise, it's a tuple, which is not None.
Won't a large number of peek calls create a never ending chain of itertools.chain objects containing references to other itertools.chain objects?
|
80

The simple answer to your question: no, there is no simple way. There are a whole lot of work-arounds.

There really shouldn't be a simple way, because of what generators are: a way to output a sequence of values without holding the sequence in memory. So there's no backward traversal.

You could write a has_next function or maybe even slap it on to a generator as a method with a fancy decorator if you wanted to.

12 Comments

fair enough, that makes sense. i knew there was no way of finding the length of a generator, but thought i might have missed a way of finding if it is initially going to generate anything at all.
I'm not sure I can agree with "there shouldn't be a simple way". There are plenty of abstractions in computer science that are designed to output a sequence of values without holding the sequence in memory, but that allow the programmer to ask whether there is another value without removing it from the "queue" if there is. There is such thing as single peek-ahead without requiring "backward traversal". That's not to say an iterator design must supply such a feature, but it sure is useful. Maybe you're objecting on the basis that the first value might change after the peek?
I'm objecting on the grounds that a typical implementation doesn't even calculate a value until it is needed. One could force the interface to do this, but that might be sub-optimal for lightweight implementations.
@S.Lott you don't need to generate the entire sequence to know if the sequence is empty or not. One element's worth of storage is sufficient - see my answer.
Too complicated a description and no visible solution for 55 upvotes!!!
|
51

A simple way is to use the optional parameter for next() which is used if the generator is exhausted (or empty). For example:

_exhausted = object() if next(some_generator, _exhausted) is _exhausted: print('generator is empty') 

10 Comments

Why objects and all that stuff? Simply: if next(itreable,-1) == -1 then the gen is empty!
@Apostolos Because next(iter([-1, -2, -3]), -1) == -1 is True. In other words, any iterable with first element equal to -1 will appear as empty using your condition.
@Apostolos In the simple case, yes, that is the solution. But it fails if you plan to create a general tool for any iterable, without constraints.
@Apostolos The object() is the extraordinary value that won't be contained in the generator.
N.B.; this is still a "peek" function and will take off one element from the generator.
|
36

Quick-dirty solution:

next(my_generator(), None) is not None

Or replace None by whatever value you know it's not in your generator.

Edit: Yes, this will skip 1 item in the generator. Sometimes, however, I check whether a generator is empty only for validation purposes, then don't really use it. Otherwise, I do something like:

def foo(self): if next(self.my_generator(), None) is None: raise Exception("Not initiated") for x in self.my_generator(): ... 

That is, this works if your generator comes from a function, as in my_generator().

6 Comments

Why this is not the best answer? In case the generator returns None?
Probably because this forces you to actually consume the generator instead of just testing if it’s empty.
It's bad because the moment you call next(generator, None) you will skip 1 item if it is available
Correct, you are going to miss 1st element of your gen and also you are going to consume your gen rather testing if its empty.
@MikkoKoho, original question (or at least the "current original," as of this comment) askes for a generator, not all generators. This answer works for a subset of all possible generators, and has an appropriate caveat which makes it work for all generators that an individual would ever use in practice.
|
24

The best approach, IMHO, would be to avoid a special test. Most times, use of a generator is the test:

thing_generated = False # Nothing is lost here. if nothing is generated, # the for block is not executed. Often, that's the only check # you need to do. This can be done in the course of doing # the work you wanted to do anyway on the generated output. for thing in my_generator(): thing_generated = True do_work(thing) 

If that's not good enough, you can still perform an explicit test. At this point, thing will contain the last value generated. If nothing was generated, it will be undefined - unless you've already defined the variable. You could check the value of thing, but that's a bit unreliable. Instead, just set a flag within the block and check it afterward:

if not thing_generated: print "Avast, ye scurvy dog!" 

4 Comments

This solution will try to consume the whole generator thus making it unusable for infinite generators.
@ViktorStískala: I don't see your point. It would be foolish to test if an infinite generator produced any results.
I wanted to point out that your solution could contain break in the for loop, because you are not processing the other results and it's useless for them to generated. range(10000000) is finite generator (Python 3), but you don't need to go through all the items to find out if it generates something.
@ViktorStískala: Understood. However, my point is this: Generally, you actually want to operate on the generator output. In my example, if nothing is generated, you now know it. Otherwise, you operate on the generated output as intended - "The use of the generator is the test". No need for special tests, or pointlessly consuming the generator output. I've edited my answer to clarify this.
11

Just fell on this thread and realized that a very simple and easy to read answer was missing:

def is_empty(generator): for item in generator: return False return True 

If we are not suppose to consume any item then we need to re-inject the first item into the generator:

def is_empty_no_side_effects(generator): try: item = next(generator) def my_generator(): yield item yield from generator return my_generator(), False except StopIteration: return (_ for _ in []), True 

Example:

>>> g=(i for i in []) >>> g,empty=is_empty_no_side_effects(g) >>> empty True >>> g=(i for i in range(10)) >>> g,empty=is_empty_no_side_effects(g) >>> empty False >>> list(g) [0, 1, 2, 3, 4, 5, 6, 7, 8, 9] 

Comments

9

Prompted by Mark Ransom, here's a class that you can use to wrap any iterator so that you can peek ahead, push values back onto the stream and check for empty. It's a simple idea with a simple implementation that I've found very handy in the past.

class Pushable: def __init__(self, iter): self.source = iter self.stored = [] def __iter__(self): return self def __bool__(self): if self.stored: return True try: self.stored.append(next(self.source)) except StopIteration: return False return True def push(self, value): self.stored.append(value) def peek(self): if self.stored: return self.stored[-1] value = next(self.source) self.stored.append(value) return value def __next__(self): if self.stored: return self.stored.pop() return next(self.source) 

1 Comment

UPDATE: I thought it was worth turning this into a PyPI package, since I have used it quite a few times in the past. Please find a slightly more elaborate version here - pypi.org/project/pushable
8

I hate to offer a second solution, especially one that I would not use myself, but, if you absolutely had to do this and to not consume the generator, as in other answers:

def do_something_with_item(item): print item empty_marker = object() try: first_item = my_generator.next() except StopIteration: print 'The generator was empty' first_item = empty_marker if first_item is not empty_marker: do_something_with_item(first_item) for item in my_generator: do_something_with_item(item) 

Now I really don't like this solution, because I believe that this is not how generators are to be used.

Comments

6

All you need to do to see if a generator is empty is to try to get the next result. Of course if you're not ready to use that result then you have to store it to return it again later.

Here's a wrapper class that can be added to an existing iterator to add an __nonzero__ test, so you can see if the generator is empty with a simple if. It can probably also be turned into a decorator.

class GenWrapper: def __init__(self, iter): self.source = iter self.stored = False def __iter__(self): return self def __nonzero__(self): if self.stored: return True try: self.value = next(self.source) self.stored = True except StopIteration: return False return True def __next__(self): # use "next" (without underscores) for Python 2.x if self.stored: self.stored = False return self.value return next(self.source) 

Here's how you'd use it:

with open(filename, 'r') as f: f = GenWrapper(f) if f: print 'Not empty' else: print 'Empty' 

Note that you can check for emptiness at any time, not just at the start of the iteration.

3 Comments

This is headed in the right direction. It should be modified to allow peeking ahead as far as you wish, storing as many results as needed. Ideally it would allow for pushing of arbitrary items onto the head of the stream. A pushable-iterator is a very useful abstraction that I often use.
@sfkleach I don't see a need to complicate this for multiple peek-ahead, it's quite useful as is and answers the question. Even though this is an old question it's still getting the occasional look, so if you want to leave your own answer someone might find it useful.
Mark's quite right that his solution answers the question, which is the key point. I should have phrased it better. What I meant was that pushable-iterators with unbounded pushback are an idiom I have found extremely useful & the implementation is arguably even simpler. As suggested I will post the variant code.
5

Sorry for the obvious approach, but the best way would be to do:

for item in my_generator: print item 

Now you have detected that the generator is empty while you are using it. Of course, item will never be displayed if the generator is empty.

This may not exactly fit in with your code, but this is what the idiom of the generator is for: iterating, so perhaps you might change your approach slightly, or not use generators at all.

4 Comments

Or... questioner could provide some hint as to why one would try to detect an empty generator?
did you mean "nothing will be displayed since generator is empty"?
S.Lott. I agree. I can't see why. But I think even if there was a reason, the problem might be better turned to use each item instead.
This does not tell the program if the generator was empty.
4

I realize that this post is 5 years old at this point, but I found it while looking for an idiomatic way of doing this, and did not see my solution posted. So for posterity:

import itertools def get_generator(): """ Returns (bool, generator) where bool is true iff the generator is not empty. """ gen = (i for i in [0, 1, 2, 3, 4]) a, b = itertools.tee(gen) try: a.next() except StopIteration: return (False, b) return (True, b) 

Of course, as I'm sure many commentators will point out, this is hacky and only works at all in certain limited situations (where the generators are side-effect free, for example). YMMV.

2 Comments

This will only call the gen generator once for each item, so side-effects are not too bad a problem. But it will store a copy of everything that has been pulled from the generator via b, but not via a, so the memory implications are similar to just running list(gen) and checking that.
It has two issues. 1. This itertool may require significant auxiliary storage (depending on how much temporary data needs to be stored). In general, if one iterator uses most or all of the data before another iterator starts, it is faster to use list() instead of tee(). 2. tee iterators are not threadsafe. A RuntimeError may be raised when using simultaneously iterators returned by the same tee() call, even if the original iterable is threadsafe.
4

I found only this solution as working for empty iterations as well.

def is_generator_empty(generator): a, b = itertools.tee(generator) try: next(a) except StopIteration: return True, b return False, b is_empty, generator = is_generator_empty(generator) 

Or if you do not want to use exception for this try to use

def is_generator_empty(generator): a, b = itertools.tee(generator) for item in a: return False, b return True, b is_empty, generator = is_generator_empty(generator) 

In the marked solution you are not able to use it for empty generators like

def get_empty_generator(): while False: yield None generator = get_empty_generator() 

Comments

3
>>> gen = (i for i in []) >>> next(gen) Traceback (most recent call last): File "<pyshell#43>", line 1, in <module> next(gen) StopIteration 

At the end of generator StopIteration is raised, since in your case end is reached immediately, exception is raised. But normally you shouldn't check for existence of next value.

another thing you can do is:

>>> gen = (i for i in []) >>> if not list(gen): print('empty generator') 

3 Comments

Which does actually consume the whole generator. Sadly, it's not clear from the question if this is desirable or undesirable behavior.
as any other way of "touching" generator, I suppose.
I realise this is old, but using 'list()' can't be the best way, if the list generated is not empty but in fact large then this is needlessly wasteful
1

If you need to know before you use the generator, then no, there is no simple way. If you can wait until after you have used the generator, there is a simple way:

was_empty = True for some_item in some_generator: was_empty = False do_something_with(some_item) if was_empty: handle_already_empty_generator_case() 

Comments

1

Simply wrap the generator with itertools.chain, put something that will represent the end of the iterable as the second iterable, then simply check for that.

Ex:

import itertools g = some_iterable eog = object() wrap_g = itertools.chain(g, [eog]) 

Now all that's left is to check for that value we appended to the end of the iterable, when you read it then that will signify the end

for value in wrap_g: if value == eog: # DING DING! We just found the last element of the iterable pass # Do something 

2 Comments

Use eog = object() instead of assuming that float('-inf') will never occur in the iterable.
@bfontaine Good idea
1

In my case I needed to know if a host of generators was populated before I passed it on to a function, which merged the items, i.e., zip(...). The solution is similar, but different enough, from the accepted answer:

Definition:

def has_items(iterable): try: return True, itertools.chain([next(iterable)], iterable) except StopIteration: return False, [] 

Usage:

def filter_empty(iterables): for iterable in iterables: itr_has_items, iterable = has_items(iterable) if itr_has_items: yield iterable def merge_iterables(iterables): populated_iterables = filter_empty(iterables) for items in zip(*populated_iterables): # Use items for each "slice" 

My particular problem has the property that the iterables are either empty or has exactly the same number of entries.

Comments

1

Use the peek function in cytoolz.

from cytoolz import peek from typing import Tuple, Iterable def is_empty_iterator(g: Iterable) -> Tuple[Iterable, bool]: try: _, g = peek(g) return g, False except StopIteration: return g, True 

The iterator returned by this function will be equivalent to the original one passed in as an argument.

Comments

1

Just to try to help with my "2 cents", I am going to describe my experience:

I have a generator that I need slicing it using itertools.islice into small generators. Then to check if my sub generators are empty or not, I just convert/consume them to a small list and I check if the list is empty or not.

For example:

from itertools import islice def generator(max_yield=10): a = 0 while True: a += 1 if a > max_yield: raise StopIteration() yield a tg = generator() label = 1 while True: itg = list(islice(tg, 3)) if not itg: # <-- I check if the list is empty or not break for i in itg: print(f'#{label} - {i}') label += 1 

Output:

#1 - 1 #1 - 2 #1 - 3 #2 - 4 #2 - 5 #2 - 6 #3 - 7 #3 - 8 #3 - 9 #4 - 10 

Maybe this is not the best approach, mainly because it consumes the generator, however it works to me.

Comments

1

Inspecting the generator before iterating over it conforms to the LBYL coding style. Another approach (EAFP) would be to iterate over it and then check whether it was empty or not.

is_empty = True for item in generator: is_empty = False do_something(item) if is_empty: print('Generator is empty') 

This approach also handles well infinite generators.

Comments

0

Here's a simple decorator which wraps the generator, so it returns None if empty. This can be useful if your code needs to know whether the generator will produce anything before looping through it.

def generator_or_none(func): """Wrap a generator function, returning None if it's empty. """ def inner(*args, **kwargs): # peek at the first item; return None if it doesn't exist try: next(func(*args, **kwargs)) except StopIteration: return None # return original generator otherwise first item will be missing return func(*args, **kwargs) return inner 

Usage:

import random @generator_or_none def random_length_generator(): for i in range(random.randint(0, 10)): yield i gen = random_length_generator() if gen is None: print('Generator is empty') 

One example where this is useful is in templating code - i.e. jinja2

{% if content_generator %} <section> <h4>Section title</h4> {% for item in content_generator %} {{ item }} {% endfor % </section> {% endif %} 

1 Comment

This calls the generator function twice, so will incur the start-up cost of the generator twice. That could be substantial if, for example, the generator function is a database query.
0

peekable from more-itertools allows checking whether it's exhausted by checking its truth value. Demo with one empty and one non-empty iterator:

from more_itertools import peekable for source in '', 'foobar': it = iter(source) if it := peekable(it): print('values:', *it) else: print('empty') 

Output:

empty values: f o o b a r 

Comments

-1

This is an old and answered question, but as no one has shown it before, here it goes:

for _ in generator: break else: print('Empty') 

You can read more here

2 Comments

But how is that useful when you do want to actually do some work with generator items? Just inserting this snippet before main code looks very dirty WA
This obviously doesn't work if generator producing more than one item.
-1

Existing answers use custom solutions to address this problem, but an off the shelf solution is available via more_itertools.peekable.

This allows the user to retain the memory efficiency of the iterator, whilst also:

  • being able to check the next available element without consuming it
  • check that the generator isn't empty
 from more_itertools import peekable def gen_fun(n: int): for i in range(n): yield i generator_object = gen_fun(5) peekable_generator_object = peekable(generator_object) if not peekable_generator_object: print("Empty Generator") 

2 Comments

Thank you for your interest in contributing to the Stack Overflow community. This question already has quite a few answers—including one that has been extensively validated by the community. Are you certain your approach hasn’t been given previously? If so, it would be useful to explain how your approach is different, under what circumstances your approach might be preferred, and/or why you think the previous answers aren’t sufficient. Can you kindly edit your answer to offer an explanation?
This answer already showed peekable.
-2

There's a very simple solution: if next(generator,-1) == -1 then the generator is empty!

7 Comments

This will consume the generator.
To recap: the question is about checking before consuming anything.
What consuming are you talking about? This is done once at start! My solution is certainly not wrong!
Although this doesn't answer the exact question as stated, I'm going to upvote it because it does deal with a common case where finding out if a generator would return anything. Quite often I find myself wanting to write something like matches = filter(lambda x: ..., my_list); return next(matches) if any_results(matches) else None. I only just learned that this can be written as matches = filter(lambda x: ..., my_list); return next(matches, None)
"My solution is certainly not wrong" is a pretty bold statement. The note about "consuming" means that by applying next to your generator, you're extracting its first value and use it to test its emptiness. The value is not stored in any variable and thus immediately discarded. When you later use the generator, the first value will be missing. Please learn more about generators.
|
-4

I solved it by using the sum function. See below for an example I used with glob.iglob (which returns a generator).

def isEmpty(): files = glob.iglob(search) if sum(1 for _ in files): return True return False 

*This will probably not work for HUGE generators but should perform nicely for smaller lists

Comments

-5

bool(generator) will return the correct result

3 Comments

easy and elegant!
I am not sure that is correct though: >>> def foo(): ... if False: ... yield 99 ... >>> bool(foo()) True >>> next(foo()) Traceback (most recent call last): File "<stdin>", line 1, in <module> StopIteration Tested on Python 3.9.10.
This is incorrect. An empty generator still gets True when converted to Boolean.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.