:vjddlmZddlmZddlmZddlmZmZm Z m Z dede e zfdZ d eeezd zd e efd Z Gd deZy )) OrderedDict)log2)Any)CallableIterableNDArraySequence operationitemsc t|dk(ryt|dk(r|dSt|dk(r||d|dSttdtt|zD]l}g}t|dz}tdt||z dD]#}|j |||||dz%|dk7r|j |d|}ntdk(sJ|dS)af Call an operation function in a cascaded pairwise way against a flat list of items. This should produce the same result as `functools.reduce` if `operation` is commutable like addition or multiplication. This may be faster for an `operation` that runs with a speed proportional to its largest input, which mesh booleans appear to. The union of a large number of small meshes appears to be "much faster" using this method. This only differs from `functools.reduce` for commutative `operation` in that it returns `None` on empty inputs rather than `functools.reduce` which raises a `TypeError`. For example on `a b c d e f g` this function would run and return: a b c d e f ab cd ef g abcd efg -> abcdefg Where `functools.reduce` would run and return: a b ab c abc d abcd e abcde f abcdef g -> abcdefg Parameters ---------- operation The function to call on pairs of items. items The flat list of items to apply operation against. rNr)lenrangeintrappend)r r _results items_modis b/workspace/moltex-experiments/.pixi/envs/default/lib/python3.12/site-packages/trimesh/iteration.pyreduce_cascadersT 5zQ UqQx Uqq58,, 3q4E ++, -JN q#e*y0!4A NN9U1XuQU|< =5 > NN59 %. w<1   1:argsNreturncg}|Dcgc]H}|Dt|dr't|ttfs|j |n|j |Jc}|Scc}w)a A less principled version of `list(itertools.chain(*args))` that accepts non-iterable values, filters `None`, and returns a list rather than yielding values. If all passed values are iterables this will return identical results to `list(itertools.chain(*args))`. Examples ---------- In [1]: list(itertools.chain([1,2], [3])) Out[1]: [1, 2, 3] In [2]: trimesh.util.chain([1,2], [3]) Out[2]: [1, 2, 3] In [3]: trimesh.util.chain([1,2], [3], 4) Out[3]: [1, 2, 3, 4] In [4]: list(itertools.chain([1,2], [3], 4)) ----> 1 list(itertools.chain([1,2], [3], 4)) TypeError: 'int' object is not iterable In [5]: trimesh.util.chain([1,2], None, 3, None, [4], [], [], 5, []) Out[5]: [1, 2, 3, 4, 5] Parameters ----------- args Will be individually checked to see if they're iterable before either being appended or extended to a flat list. Returns ---------- chained The values in a flat list. __iter__)hasattr isinstancestrbytesextendr)rchainedas rchainr&PsoVG  A = Az ":a#u+F q ^^A    NsA AcZeZdZdZfdZfdZdZexZxZxZ Z dfd Z de fdZ xZS) IndexedDicta An append-only `OrderedDict` which knows what position a key was inserted at. Useful anywhere values are referenced by *position* but keyed by content so duplicates are only stored once: the only other spelling is `list(d.keys()).index(key)`, which allocates every key and scans it, i.e. quadratic. Looking up the position of all `n` keys once each: n list(keys()).index() this class 2000 0.061s 0.00007s 4000 0.262s 0.00012s 8000 1.135s 0.00026s Removing or reordering a key would shift the position of every key after it, so `__delitem__`, `pop`, `popitem`, and `move_to_end` raise: the supported way to remove is `clear` followed by `update`. Examples ---------- In [1]: IndexedDict({"a": 1, "b": 2, "c": 3}).index("c") Out[1]: 2 c2i|_t||i|yN) _positionsuper__init__)selfrkwargs __class__s rr-zIndexedDict.__init__s $)&)rc^||vrt||j|<t| ||yr*)rr+r, __setitem__)r.keyvaluer0s rr2zIndexedDict.__setitem__s+ d?"%d)DNN3  C'rctd)Nz6`IndexedDict` is append-only: use `clear` and `update`) TypeError)r.rr/s r _forbiddenzIndexedDict._forbiddensPQQrrcVt||jjyr*)r,clearr+)r.r0s rr9zIndexedDict.clears   rc |j|S)zJ Which position in insertion order was `key` inserted at. )r+)r.r3s rindexzIndexedDict.indexs~~c""r)rN)__name__ __module__ __qualname____doc__r-r2r7 __delitem__poppopitem move_to_endr9rr; __classcell__)r0s@rr(r(s?6* ( R1;:K:#:+#C#rr() collectionsrmathrtypingrtypedrrr r rlistr&r(rrrKsb#88EhEx'/AEP4#$t+4S 4n4#+4#r