+- Python Forum (https://python-forum.io)
+-- Forum: Python Coding (https://python-forum.io/forum-7.html)
+--- Forum: General Coding Help (https://python-forum.io/forum-8.html)
+--- Thread: compacting a mutable sequence (/thread-7713.html)
compacting a mutable sequence - Skaperen - Jan-22-2018
i am looking for some already existing implementation of this logic i am ready to implement for myself, if necessary. i want to "compact" a mutable sequence by removing duplicates, keeping the order of both the unique items and each first item of duplicates. converting to a set loses the original order, so it is not an option to convert to a set and convert back, but maybe a set can be used to help detect duplicates in an implementation. anyone know of an existing implementation?
RE: compacting a mutable sequence - Larz60+ - Jan-22-2018
there's zlib:
import zlib
class TryZlibBz2:
def __init__(self):
your_text = str.encode('i am looking for some already existing implementation of this logic i am ready to implement for myself, if necessary. i want to "compact" a mutable sequence by removing duplicates, keeping the order of both the unique items and each first item of duplicates. converting to a set loses the original order, so it is not an option to convert to a set and convert back, but maybe a set can be used to help detect duplicates in an implementation. anyone know of an existing implementation?')
compressed = self.zlib_encode(your_text)
print(f'orig len: {len(your_text)}, len compressed: {len(compressed)}')
print(f'\n{self.zlib_decode(compressed).decode()}')
def zlib_encode(self, text):
return zlib.compress(text)
def zlib_decode(self, enctext):
return zlib.decompress(enctext)
if __name__ == '__main__':
TryZlibBz2()results:Output:orig len: 484, len compressed: 263
i am looking for some already existing implementation of this logic i am ready to implement for myself, if necessary. i want to "compact" a mutable sequence by removing duplicates, keeping the order of both the unique items and each first item of duplicates. converting to a set loses the original order, so it is not an option to convert to a set and convert back, but maybe a set can be used to help detect duplicates in an implementation. anyone know of an existing implementation?RE: compacting a mutable sequence - Skaperen - Jan-22-2018
sorry i was not clear. i did not mean compression. maybe deduplicate or unduplicate would have been a better choice of terminology. and i expected to do this directly to mutable sequences such as lists.
[7,4,3,4,9,4,'foo',3,1,7,5] -> [7,4,3,9,'foo',1,5]
the duplicate items are removed and those items that stay keep their original order.
RE: compacting a mutable sequence - Gribouillis - Jan-22-2018
Don't you know
unique_everseen() ? Look in itertools documentation.
RE: compacting a mutable sequence - Skaperen - Jan-22-2018
(Jan-22-2018, 06:29 AM)Gribouillis Wrote: Don't you know unique_everseen() ? Look in itertools documentation.no ... i do not know of it. i do now. thanks! i suspected this might exist but had no ideal what name.
but...
itertools.unique_everseen() is limited to hashable items. i don't know if i will need to support non-hashable or not. so i still might need to implement my own, in which case a deep compare would be handy (have not looked for one,yet)
RE: compacting a mutable sequence - Gribouillis - Jan-22-2018
(Jan-22-2018, 06:46 AM)Skaperen Wrote: i don't know if i will need to support non-hashable or not.If you don't know, don't implement it! More seriously, there is a
key parameter in unique_everseen. It should be able to handle many non hashable cases because a non hashable item can have a hashable key.
RE: compacting a mutable sequence - Skaperen - Jan-23-2018
if it turns out that the lists i need to compact might have non-hashable objects then i will simply need to use a less efficient method to keep the memory of what objects have been seen. what i am thinking about is to start running the hashable way (keeping seen objects in a set) while hashing each object, and if any object is ever found to be unhashable, then switch over to the method that handles unhashable objexts (keeping seen objects in a list that is scanned linearly). that way i can have the O(n*log(n)) performance of keeping a set, but still be able to work with a mutable sequence of unhashable objects.