Question

I was looking for a way to create multiple ad-hoc copies of a dictionary to hold some “evolutionary states”, with just slight generational deviations and found this little prototype-dict:

class ptdict(dict):
    def asprototype(self, arg=None, **kwargs):
        clone = self.__class__(self.copy())
        if isinstance(arg, (dict, ptdict)):
            clone.update(arg)
        clone.update(self.__class__(kwargs))        
        return clone

Basically i want smth. like:

generation0 = dict(propertyA="X", propertyB="Y")
generations = [generation0]

while not endofevolution():
    # prev. generation = template for next generation:
    nextgen = generations[-1].mutate(propertyB="Z", propertyC="NEW")  
    generations.append(nextgen)

and so on.

I was wondering, if the author of this class and me were missing something, because i just can’t imagine, that there’s no standard-library approach for this. But neither the collections nor the itertools seemed to provide a similar simple approach.

Can something like this be accomplished with itertools.tee?

Update:
It’s not a question of copy & update, because, that’s exactly what this ptdict is doing. But using update doesn’t return a dict, which ptdict does, so i can for example chain results or do in-place tests, which would enhance readability quite a bit. (My provided example is maybe a bit to trivial, but i didn’t want to confuse with big matrices.)

I apologise for not having been precise enough. Maybe the following example clarifies why i’m interested in getting a dictionary with a single copy/update-step:

nextgen = nextgen.mutate(inject_mutagen("A")) if nextgen.mutate(inject_mutagen("A")).get("alive") else nextgen.mutate(inject_mutagen("C"))

Answer 1

I guess you’re looking for something like this:

first = {'x':1, 'y':100, 'foo':'bar'}
second = dict(first, x=2, y=200) # {'y': 200, 'x': 2, 'foo': 'bar'}

See dict