另一个UnboundLocalError：在赋值问题之前引用的局部变量

Question

I stumbled upon a situation that shutters my understanding of Pythons variable scope.

Here is the code:

transaction_id = None


def parseFileContent(hostID,marketID, content, writeToDB=False):
    features = _buildObjects(StringIO.StringIO(content))

    for feature in features:
        featureID =  adapter.addFeature(feature.name,boris)
        print transaction_id #breaks here UnboundLocalError: local variable 'transaction_id' referenced before assignment

        transaction_id = adapter.addFeatureTransactionToQueue(featureID, result[0], result[1], Command.ADD, boris, trans_id = transaction_id)

If I replace last line with

       adapter.addFeatureTransactionToQueue(featureID, result[0], result[1], Command.ADD, boris, trans_id = transaction_id)

Everything works. I need to understand what python dislikes about me printing the value in the first scenario.

Answer 1

The Python compiler marks a name as local to a function if you assign to it . Your last line assigns to transaction_id so it is seen as a local name, not a global.

You need to tell the compiler explicitly that transaction_id is a global, by using the global keyword inside the function:

def parseFileContent(hostID,marketID, content, writeToDB=False):
    global transaction_id

If there is no assignment, a name is considered non-local instead and you do not need to mark it.

Answer 2

Since Python doesn't have variable declarations, it needs another way to tell what's local to what scope. It does so by defining that anything a function assigns to is local to that function unless overridden by a global declaration. Thus, assigning to transaction_id makes it a local, and the print tries to print the unassigned local and fails.

Answer 3

When you assign to a name inside a function, it's a local name. Python will only ever treat the name as local inside that function, even if you have a global variable with the same name. This is different from some other languages, where if a local is not found, a global with the same name will be used if it exists.

You should avoid global variables. There is no reason for your transaction_id to be global in this case. So just put the None assignment inside the function. If you want access to the last transaction ID after the function executes, simply return it from the function.

def parseFileContent(hostID,marketID, content, writeToDB=False):
    features = _buildObjects(StringIO.StringIO(content))
    transaction_id = None     # <----------------------

    for feature in features:
        featureID =  adapter.addFeature(feature.name,boris)
        print transaction_id  # now works

        transaction_id = adapter.addFeatureTransactionToQueue(featureID, 
            result[0], result[1], Command.ADD, boris, trans_id=transaction_id)

     return transaction_id    # <----------------------