How to turn a list into nested dict in Python
Solution 1
X = [['A', 'B', 'C'], ['A', 'B', 'D'],['W','X'],['W','Y','Z']]
d = {}
for path in X:
current_level = d
for part in path:
if part not in current_level:
current_level[part] = {}
current_level = current_level[part]
This leaves us with d containing {'A': {'B': {'C': {}, 'D': {}}}, 'W': {'Y': {'Z': {}}, 'X': {}}}
. Any item containing an empty dictionary is either a file or an empty directory.
Solution 2
Assuming that {'C', 'D'}
means set(['C', 'D'])
and your Python version supports dict comprehension
and set comprehension
, here's an ugly but working solution:
>>> tr = [[1, 2, 3], [1, 2, 4], [5, 6, 7]]
>>> {a[0]: {b[1]: {c[2] for c in [y for y in tr if y[1] == b[1]]} for b in [x for x in tr if x[0] == a[0]]} for a in tr}
{1: {2: set([3, 4])}, 5: {6: set([7])}}
As for your example:
>>> X = [['A', 'B', 'C'], ['A', 'B', 'D']]
>>> {a[0]: {b[1]: {c[2] for c in [y for y in X if y[1] == b[1]]} for b in [x for x in X if x[0] == a[0]]} for a in X}
{'A': {'B': set(['C', 'D'])}}
But please don't use it in a real-world application :)
UPDATE: here's one that works with arbitrary depths:
>>> def todict(lst, d=0):
... print lst, d
... if d > len(lst):
... return {}
... return {a[d]: todict([x for x in X if x[d] == a[d]], d+1) for a in lst}
...
>>> todict(X)
{'A': {'B': {'C': {}, 'D': {}}}}
Solution 3
There is a logical inconsistency in your problem statement. If you really want
['xyz/123/file.txt', 'abc/456/otherfile.txt']
to be changed to
{'xyz': {'123': 'file.txt}, 'abc': {'456': 'otherfile.txt'}}
Then you have to answer how a path 'abc.txt' with no leading folder would be inserted into this data structure. Would the top-level dictionary key be the empty string ''
?
Solution 4
I got asked about this question on twitter and came up with this slick solution using functional programming which I figure I might as well share here.
from functools import reduce
X = [['A', 'B', 'C'], ['A', 'B', 'D']]
Y = [reduce(lambda x, y: {y:x}, Y[::-1]) for Y in X]
which returns:
[{'A': {'B': 'C'}}, {'A': {'B': 'D'}}]
as desired.
For the simpler problem where you have one list that you want to represent as a dict with nested keys, this will suffice:
from functools import reduce
X = ['A', 'B', 'C']
reduce(lambda x, y: {y:x}, X[::-1])
which returns:
{'A': {'B': 'C'}}
Solution 5
This should be pretty close to what you need:
def path_to_dict(path):
parts = path.split('/')
def pack(parts):
if len(parts) == 1:
return parts
elif len(parts):
return {parts[0]: pack(parts[1:])}
return parts
return pack(parts)
if __name__ == '__main__':
paths = ['xyz/123/file.txt', 'abc/456/otherfile.txt']
for path in paths:
print '%s -> %s' % (path, path_to_dict(path))
Results in:
xyz/123/file.txt -> {'xyz': {'123': ['file.txt']}}
abc/456/otherfile.txt -> {'abc': {'456': ['otherfile.txt']}}
Dmitry Grinberg
Updated on June 05, 2022Comments
-
Dmitry Grinberg almost 2 years
Need to turn x:
X = [['A', 'B', 'C'], ['A', 'B', 'D']]
Into Y:
Y = {'A': {'B': {'C','D'}}}
More specifically, I need to create a tree of folders and files from a list of absolute paths, which looks like this:
paths = ['xyz/123/file.txt', 'abc/456/otherfile.txt']
where, each path is
split("/")
, as per['A', 'B', 'C']
in the pseudo example.As this represents files and folders, obviously, on the same level (index of the array) same name strings can't repeat.
-
FogleBird over 12 yearsWhat's this?
{'C','D'}
-
Steven Rumbalski over 12 yearsWhat problem are you trying to solve with the nested dict?
-
Steven Rumbalski over 12 yearsWhat if folder A contains a directory B and also a file X? How should that be represented?
-
FogleBird over 12 yearsWould this be acceptable?
{'A': {'B': {'C': {}, 'D': {}}}}
-
Dmitry Grinberg over 12 yearsmore consistent, Y = {'A': {'B': {'C':{},'D':{}}}}
-
Dmitry Grinberg over 12 yearsSteven, I'm actually trying to create a page with nested ul/li blocks, to render a javascript tree. the way content is saved, though, it's just a list of records, with URL property, that represents an absolute path. I can't just show a flat list, so need to show it as if it's a list of nested directories with files in them.
-
Karl Knechtel over 12 yearsWhere does the input data come from? If you want to iterate over a directory structure, the normal approach is to just use
os.walk
.
-
-
agf over 12 yearsAlso doesn't work for arbitrary depth, which is probably a requirement.
-
Attila O. over 12 years@agf true - I didn't think that was a requirement, it probably is though. Should I come up with an even more horrible looking one-liner? :)
-
agf over 12 yearsHah, that would just get scary. I think Matt's got the idea -- it has to be done recursively.
-
agf over 12 yearsHow do you join all the dictionaries, like if you have other paths like
xyz/123/b.a
andabc/sss/ooo/1.a
that overlap with those? -
Dmitry Grinberg over 12 yearsif there is a file called abc.txt at top level, it'll be a peer with 'xyz' and 'abc', as in {'xyz':{...}, 'abc':{...}, 'abc.txt:{}} . with empty {}, which would indicate it's a leaf
-
Dmitry Grinberg over 12 yearsit doesn't matter if something "looks" like a file or folder. last one will be a file, rest are folders. this is not real file system, btw.
-
Steven Rumbalski over 12 yearsWhat if
'xyz'
contains directories and files? -
Dmitry Grinberg over 12 yearsMatt, I was about to ask same question as agf. This far i got myself. I actually need one dict merged. there will be overlaps, so I can't just replace nodes wholesale, need to actually merge at every level.
-
Matt Williamson over 12 yearsDeeply merge dicts: appdelegateinc.com/blog/2011/01/12/…
-
Tyler Hilbert about 7 yearsWhere does the current_level get added to d?
-
tsveti_iko over 5 years@Tyler Hilbert the
current_level
is thed
! The expressioncurrent_level = d
doesn't make a copy ofd
, it just creates a new reference tod
namedcurrent_level
;) The code above wouldn't work ifcurrent_level
was a copy, created like this:current_level = d[:]
or like this:current_level = copy.deepcopy(d)
, but since it's not - it merely works! -
ThePhi almost 4 years@tsveti_iko If it's a reference, why doesn't the line
current_level = current_level[part]
modify onlycurrent_level
and notd
? I've checked in the debugger, and after this stage (for ex. when we are at the 1st elementA
), we'll get:current_level = {}
whiled = {'A': {}}
. Thanks! -
SpectreVert over 2 yearsabove link is dead redirects to spam