ChatGPT解决这个技术问题 Extra ChatGPT

What is the intended use of the optional "else" clause of the "try" statement in Python?

What is the intended use of the optional else clause of the try statement?

Most answers seem to concentrate on why we can't just put the material in the else clause in the try clause itself. The question stackoverflow.com/questions/3996329 specifically asks why the else clause code cannot go after the try block itself, and that question is dupped to this one, but I do not see a clear reply to that question here. I feel stackoverflow.com/a/3996378/1503120 excellently answers that question. I have also tried to elucidate the various significance of the various clauses at stackoverflow.com/a/22579805/1503120.
You want something to occur if the exception doesn't trigger, before the final cleanup, that isn't ever supposed to itself trigger the same exception handling.
After forgetting what else does in try/else and for/else numerous times, I have mentally aliased it with noexcept and nobreak in those respective contexts. Personally I find it to be such an unfortunate overloading of the word that I try to avoid using it if I can, as it forces people reading code to wonder "what does this thing do again?" Usually a flag, a continue or break statement can relay what I'm trying to express with few extra lines, but certainly more clarity (if the popularity of the question is any indication).
It is better to name it except-else instead of try-else.

v
vinzee

The statements in the else block are executed if execution falls off the bottom of the try - if there was no exception. Honestly, I've never found a need.

However, Handling Exceptions notes:

The use of the else clause is better than adding additional code to the try clause because it avoids accidentally catching an exception that wasn’t raised by the code being protected by the try ... except statement.

So, if you have a method that could, for example, throw an IOError, and you want to catch exceptions it raises, but there's something else you want to do if the first operation succeeds, and you don't want to catch an IOError from that operation, you might write something like this:

try:
    operation_that_can_throw_ioerror()
except IOError:
    handle_the_exception_somehow()
else:
    # we don't want to catch the IOError if it's raised
    another_operation_that_can_throw_ioerror()
finally:
    something_we_always_need_to_do()

If you just put another_operation_that_can_throw_ioerror() after operation_that_can_throw_ioerror, the except would catch the second call's errors. And if you put it after the whole try block, it'll always be run, and not until after the finally. The else lets you make sure

the second operation's only run if there's no exception, it's run before the finally block, and any IOErrors it raises aren't caught here


Also keep in mind that variables used in the try-block CAN be used in the else-block, so you should alway consider using this variant if you don't expect more exceptions in the else-block
That doesn't matter, because try-scoped variables are seen outside of the try whether there's an else or not.
There's no such thing as a "try-scoped variable". In Python, variable scopes are established only by modules, functions and comprehensions, not control structures.
The else clause lets you write code that only makes sense if an exception wasn't thrown; the except clause can simply pass. If you put the logic in the try block, you risk silently hiding bugs in your code. Never squash exceptions you didn't expect.
it is not clear from this answer what the "falls off the bottom" means - not only does this happen because of an exception but also, because of a return, continue or break.
x
xenosoz

There is one big reason to use else - style and readability. It's generally a good idea to keep code that can cause exceptions near the code that deals with them. For example, compare these:

try:
    from EasyDialogs import AskPassword
    # 20 other lines
    getpass = AskPassword
except ImportError:
    getpass = default_getpass

and

try:
    from EasyDialogs import AskPassword
except ImportError:
    getpass = default_getpass
else:
    # 20 other lines
    getpass = AskPassword

The second one is good when the except can't return early, or re-throw the exception. If possible, I would have written:

try:
    from EasyDialogs import AskPassword
except ImportError:
    getpass = default_getpass
    return False  # or throw Exception('something more descriptive')

# 20 other lines
getpass = AskPassword

Note: Answer copied from recently-posted duplicate here, hence all this "AskPassword" stuff.


R
Russia Must Remove Putin

Python try-else What is the intended use of the optional else clause of the try statement?

The intended use is to have a context for more code to run if there were no exceptions where it was expected to be handled.

This context avoids accidentally handling errors you did not expect.

But it's important to understand the precise conditions that cause the else clause to run, because return, continue, and break can interrupt the control flow to else.

In Summary

The else statement runs if there are no exceptions and if not interrupted by a return, continue, or break statement.

The other answers miss that last part.

From the docs:

The optional else clause is executed if and when control flows off the end of the try clause.*

(Bolding added.) And the footnote reads:

*Currently, control “flows off the end” except in the case of an exception or the execution of a return, continue, or break statement.

It does require at least one preceding except clause (see the grammar). So it really isn't "try-else," it's "try-except-else(-finally)," with the else (and finally) being optional.

The Python Tutorial elaborates on the intended usage:

The try ... except statement has an optional else clause, which, when present, must follow all except clauses. It is useful for code that must be executed if the try clause does not raise an exception. For example: for arg in sys.argv[1:]: try: f = open(arg, 'r') except IOError: print 'cannot open', arg else: print arg, 'has', len(f.readlines()), 'lines' f.close() The use of the else clause is better than adding additional code to the try clause because it avoids accidentally catching an exception that wasn’t raised by the code being protected by the try ... except statement.

Example differentiating else versus code following the try block

If you handle an error, the else block will not run. For example:

def handle_error():
    try:
        raise RuntimeError('oops!')
    except RuntimeError as error:
        print('handled a RuntimeError, no big deal.')
    else:
        print('if this prints, we had no error!') # won't print!
    print('And now we have left the try block!')  # will print!

And now,

>>> handle_error()
handled a RuntimeError, no big deal.
And now we have left the try block!

main point from the docs: "The use of the else clause is better than adding additional code to the try clause because it avoids accidentally catching an exception that wasn’t raised by the code being protected by the try ... except statement."
I think this is the best explanation. Thanks Aaron.
D
Darius Bacon

One use: test some code that should raise an exception.

try:
    this_should_raise_TypeError()
except TypeError:
    pass
except:
    assert False, "Raised the wrong exception type"
else:
    assert False, "Didn't raise any exception"

(This code should be abstracted into a more generic test in practice.)


A
Alice Purcell

Try-except-else is great for combining the EAFP pattern with duck-typing:

try:
  cs = x.cleanupSet
except AttributeError:
  pass
else:
  for v in cs:
    v.cleanup()

You might thing this naïve code is fine:

try:
  for v in x.cleanupSet:
    v.clenaup()
except AttributeError:
  pass

This is a great way of accidentally hiding severe bugs in your code. I typo-ed cleanup there, but the AttributeError that would let me know is being swallowed. Worse, what if I'd written it correctly, but the cleanup method was occasionally being passed a user type that had a misnamed attribute, causing it to silently fail half-way through and leave a file unclosed? Good luck debugging that one.


A
Asclepius

I find it really useful when you've got cleanup to do that has to be done even if there's an exception:

try:
    data = something_that_can_go_wrong()
except Exception as e: # yes, I know that's a bad way to do it...
    handle_exception(e)
else:
    do_stuff(data)
finally:
    clean_up()

A
Asclepius

Even though you can't think of a use of it right now, you can bet there has to be a use for it. Here is an unimaginative sample:

With else:

a = [1,2,3]
try:
    something = a[2]
except:
    print "out of bounds"
else:
    print something

Without else:

try:
    something = a[2]
except:
    print "out of bounds"

if "something" in locals():
    print something

Here you have the variable something defined if no error is thrown. You can remove this outside the try block, but then it requires some messy detection if a variable is defined.


What's wrong with something = a[2]; print something inside the try: block?
@ S.Lott nothing, but what if someone is sending you a list, and you don't want to display the data if its not long enough because it is probably corrupted?
S. Lott: 'print something' could raise a different exception that you don't want to intercept.
I don't see the difference. If I get an out of bounds exception, it prints "out of bounds". Got that. If I get some other exception, it's uncaught by this block of code. If I get no exception, the behavior is to print the value of something, which is a[2]. I don't see what the else does in this example.
The value of 'something', when printed, might raise the error in its __str__() method. While that value is actually just 2 in this example, you might just as well point out that there is no out-of-bounds exception here either.
i
itsadok

There's a nice example of try-else in PEP 380. Basically, it comes down to doing different exception handling in different parts of the algorithm.

It's something like this:

try:
    do_init_stuff()
except:
    handle_init_suff_execption()
else:
    try:
        do_middle_stuff()
    except:
        handle_middle_stuff_exception()

This allows you to write the exception handling code nearer to where the exception occurs.


f
fedorqui

From Errors and Exceptions # Handling exceptions - docs.python.org

The try ... except statement has an optional else clause, which, when present, must follow all except clauses. It is useful for code that must be executed if the try clause does not raise an exception. For example: for arg in sys.argv[1:]: try: f = open(arg, 'r') except IOError: print 'cannot open', arg else: print arg, 'has', len(f.readlines()), 'lines' f.close() The use of the else clause is better than adding additional code to the try clause because it avoids accidentally catching an exception that wasn’t raised by the code being protected by the try ... except statement.


S
Shivam Bharadwaj
try:
    statements # statements that can raise exceptions
except:
    statements # statements that will be executed to handle exceptions
else:
    statements # statements that will be executed if there is no exception

Example :

try:
    age=int(input('Enter your age: '))
except:
    print ('You have entered an invalid value.')
else:
    if age <= 21:
        print('You are not allowed to enter, you are too young.')
    else:
        print('Welcome, you are old enough.')

The Output :

>>> 
Enter your age: a
You have entered an invalid value.
>>> RESTART
>>> 
Enter your age: 25
Welcome, you are old enough.
>>>RESTART
>>> 
Enter your age: 13
You are not allowed to enter, you are too young.
>>>

Copied from : https://geek-university.com/python/the-try-except-else-statements/


A
Asclepius

Looking at Python reference it seems that else is executed after try when there's no exception. The optional else clause is executed if and when control flows off the end of the try clause. 2 Exceptions in the else clause are not handled by the preceding except clauses.

Dive into python has an example where, if I understand correctly, in try block they try to import a module, when that fails you get exception and bind default but when it works you have an option to go into else block and bind what is required (see link for the example and explanation).

If you tried to do work in catch block it might throw another exception - I guess that's where the else block comes handy.


"Exceptions in the else clause are not handled by the preceding except clauses." That is the useful part. Thank you.
"The optional else clause is executed if and when control flows off the end of the try clause" is another difference, since you can return out of the try block.
t
tzot

That's it. The 'else' block of a try-except clause exists for code that runs when (and only when) the tried operation succeeds. It can be used, and it can be abused.

try:
    fp= open("configuration_file", "rb")
except EnvironmentError:
    confdata= '' # it's ok if the file can't be opened
else:
    confdata= fp.read()
    fp.close()

# your code continues here
# working with (possibly empty) confdata

Personally, I like it and use it when appropriate. It semantically groups statements.


v
vadimbog

I would add another use case that seems straight forward when handling DB sessions:

    # getting a DB connection 
    conn = db.engine.connect()

    # and binding to a DB session
    session = db.get_session(bind=conn)

    try:
        # we build the query to DB
        q = session.query(MyTable).filter(MyTable.col1 == 'query_val')

        # i.e retrieve one row
        data_set = q.one_or_none()

        # return results
        return [{'col1': data_set.col1, 'col2': data_set.col2, ...}]

    except:
        # here we make sure to rollback the transaction, 
        # handy when we update stuff into DB
        session.rollback()
        raise

    else:
        # when no errors then we can commit DB changes
        session.commit()

    finally:
        # and finally we can close the session
        session.close()

To me seems like it's only useful if you're using finally, else you can just put the else code outside the try scope
D
DevPlayer

Perhaps a use might be:

#debug = []

def debuglog(text, obj=None):
    " Simple little logger. "
    try:
        debug   # does global exist?
    except NameError:
        pass    # if not, don't even bother displaying
    except:
        print('Unknown cause. Debug debuglog().')
    else:
        # debug does exist.
        # Now test if you want to log this debug message
        # from caller "obj"
        try:
            if obj in debug:
                print(text)     # stdout
        except TypeError:
            print('The global "debug" flag should be an iterable.')
        except:
            print('Unknown cause. Debug debuglog().')

def myfunc():
    debuglog('Made it to myfunc()', myfunc)

debug = [myfunc,]
myfunc()

Maybe this will lead you too a use.


s
sirlark

I have found the try: ... else: construct useful in the situation where you are running database queries and logging the results of those queries to a separate database of the same flavour/type. Let's say I have lots of worker threads all handling database queries submitted to a queue

#in a long running loop
try:
    query = queue.get()
    conn = connect_to_db(<main db>)
    curs = conn.cursor()
    try:
        curs.execute("<some query on user input that may fail even if sanitized">)
    except DBError:
        logconn = connect_to_db(<logging db>)
        logcurs = logconn.cursor()
        logcurs.execute("<update in DB log with record of failed query")
        logcurs.close()
        logconn.close()
    else:

        #we can't put this in main try block because an error connecting
        #to the logging DB would be indistinguishable from an error in 
        #the mainquery 

        #We can't put this after the whole try: except: finally: block
        #because then we don't know if the query was successful or not

        logconn = connect_to_db(<logging db>)
        logcurs = logconn.cursor()
        logcurs.execute("<update in DB log with record of successful query")
        logcurs.close()
        logconn.close()
        #do something in response to successful query
except DBError:
    #This DBError is because of a problem with the logging database, but 
    #we can't let that crash the whole thread over what might be a
    #temporary network glitch
finally:
    curs.close()
    conn.close()
    #other cleanup if necessary like telling the queue the task is finished

Of course if you can distinguish between the possible exceptions that might be thrown, you don't have to use this, but if code reacting to a successful piece of code might throw the same exception as the successful piece, and you can't just let the second possible exception go, or return immediately on success (which would kill the thread in my case), then this does come in handy.


A
Asclepius

Here is another place where I like to use this pattern:

 while data in items:
     try
        data = json.loads(data)
     except ValueError as e:
        log error
     else:
        # work on the `data`

You can just use continue instead – the "break out early" pattern. This allows you to drop the "else" clause and its indentation, making the code easier to read.
W
Wesley

An else block can often exist to complement functionality that occurs in every except block.

try:
    test_consistency(valuable_data)
except Except1:
    inconsistency_type = 1
except Except2:
    inconsistency_type = 2
except:
    # Something else is wrong
    raise
else:
    inconsistency_type = 0

"""
Process each individual inconsistency down here instead of
inside the except blocks. Use 0 to mean no inconsistency.
"""

In this case, inconsistency_type is set in each except block, so that behaviour is complemented in the no-error case in else.

Of course, I'm describing this as a pattern that may turn up in your own code someday. In this specific case, you just set inconsistency_type to 0 before the try block anyway.


N
NeoWang

One of the use scenarios I can think of is unpredictable exceptions, which can be circumvented if you try again. For instance, when the operations in try block involves random numbers:

while True:
    try:
        r = random.random()
        some_operation_that_fails_for_specific_r(r)
    except Exception:
        continue
    else:
        break

But if the exception can be predicted, you should always choose validation beforehand over an exception. However, not everything can be predicted, so this code pattern has its place.


You can do this putting the break inside the try at the end, which is cleaner IMO, and you don't need the else. Also the continue is not really needed, you can just pass.
N
Nick

I have found else useful for dealing with a possibly incorrect config file:

try:
    value, unit = cfg['lock'].split()
except ValueError:
    msg = 'lock monitoring config must consist of two words separated by white space'
    self.log('warn', msg)
else:
     # get on with lock monitoring if config is ok

An exception reading the lock config disables lock monitoring and ValueErrors log a helpful warning message.


G
Gene

Suppose your programming logic depends on whether a dictionary has an entry with a given key. You can test the result of dict.get(key) using if... else... construct, or you can do:

try:
    val = dic[key]
except KeyError:
    do_some_stuff()
else:
    do_some_stuff_with_val(val)

G
Gulzar

One could use this construct for handling exceptions in a common way within the finally clause, while doing something else when there's no exception:

class TooManyRetries(RuntimeError):
    pass


n_tries = 0
max_retries = 2
while True:
    try:
        n_tries += 1
        if n_tries >= max_retries:
            raise TooManyRetries
        fail_prone_operation()
    except Exception1 as ex:
        # handle1
    except Exception2 as ex:
        # handle2
    except Exception3 as ex:
        # handle3
    except TooManyRetries as ex:
        raise
    else: # No exception
        n_tries = 0
    finally:
        common_restore_state()
        continue

    

S
S.Lott

The else: block is confusing and (nearly) useless. It's also part of the for and while statements.

Actually, even on an if-statement, the else: can be abused in truly terrible ways creating bugs that are very hard to find.

Consider this.

   if a < 10:
       # condition stated explicitly
   elif a > 10 and b < 10:
       # condition confusing but at least explicit
   else:
       # Exactly what is true here?
       # Can be hard to reason out what condition is true

Think twice about else:. It is generally a problem. Avoid it except in an if-statement and even then consider documenting the else- condition to make it explicit.


I would disagree with this one. In "if-elif" block, "else" is used as "default" would be used in "case" block of C language. It is always recommended to handle "default" case even if you think you've covered all cases in various conditions.
@Josip: used as a "default" can be confusing. The issue is to clearly define the condition that is this "default". A poorly-defined default condition can be the root cause of buggy behavior. Else can be a cause of confusion. It should be thought through very carefully in all cases, not just try, for and while, but if as well.
Well, the above code is totally abstract and doesn't do anything meaningful, so yea - no wonder it is confusing.
@S.Lott "It would reduce bugginess" - and my point is that this is false. I think we just have a genuine difference in opinions. Bad programmers always find ways to write buggy programs. Always. Good programmers always seek good practices and can write good code in just about any language. Eliminating useful constructs just gives less power to the good programmers while not particularily helping out the bad ones as those are capable of inventing an infinite number of ways to f**k things up.
Consider: if x > 0: return "yes" and if x <= 0: return "no". Now a person comes and changes one of the conditions to say x > 1 but forgets to change the other one. How is that reducing number of bugs that would be committed. if else clauses are sometimes many lines apart. DRY is a good practice, much more often than not, really. (sorry for the double post).