a j)@sdZddlZddlZddlZddlZddlZddlZddlZejZ Gddde Z Gddde Z Gddde Z Gd d d e Zdd dZdS)zb Fast cryptographic hash of Python objects, with a special case for fast hashing of numpy arrays. Nc@seZdZdZddZdS)_ConsistentSetz\Class used to ensure the hash of Sets is preserved whatever the order of its items. c Cs@zt||_Wn,ttjfy:tdd|D|_Yn0dS)Ncss|]}t|VqdSNhash).0er4E:\soccer\ml_env\Lib\site-packages\joblib/hashing.py %z*_ConsistentSet.__init__..)sorted _sequence TypeErrordecimalInvalidOperation)selfZ set_sequencerrr __init__sz_ConsistentSet.__init__N__name__ __module__ __qualname____doc__rrrrr rsrc@seZdZdZddZdS)_MyHashz5Class used to hash objects that won't normally picklecGs ||_dSr)args)rrrrr r+sz_MyHash.__init__Nrrrrr r(src@seZdZdZdddZdddZdd Zd d Zd ej fd dZ e j Z e e ee<e e ee<e e ee <e e eej<ddZddZee ee<d S)HasherzA subclass of pickler, to do cryptographic hashing, rather than pickling. This is used to produce a unique hash of the given Python object that is not necessarily cryptographically secure. md5cCs4t|_d}tj||j|dtj|dd|_dS)N)protocolF)usedforsecurity)ioBytesIOstreamPicklerrhashlibnew_hash)r hash_namerrrr r5s zHasher.__init__Tc Csxz||Wn@tjyN}z&|jd||ff7_WYd}~n d}~00|j}|j||rt|jSdS)Nz"PicklingError while hashing %r: %r) dumppickle PicklingErrorrr!getvaluer%update hexdigest)robjZ return_digestrdumpsrrr r>s  z Hasher.hashcCst|tjtijfr|t|dr*|jj}n|j}|j}t|tt urTt ||j}n(|durht ||}n|jj }t |||}t ||dS)N__func__) isinstancetypes MethodTypetypepophasattrr/r__self__r(r __class__r"save)rr- func_nameinstclsrrr r8Is    z Hasher.savecCs"t|ttfrdSt||dSr)r0bytesstrr"memoize)rr-rrr r>\szHasher.memoizeNcCst||d}|d=ztj||fi|Wnntjytj||fi|t|dd}|dkr|}|durt|j}tj|}t ||st |||Yn0dS)N)namepackr@r__main__) dictr" save_globalr(r)getattrrsysmodulesr5setattr)rr-r?r@kwargsmoduleZmy_namemodrrr rCgs    zHasher.save_globalc Gs^z tj|tt|g|RWn8tyXtj|ttdd|Dg|RYn0dS)Ncss|]\}}t||fVqdSrr)rkvrrr r r z)Hasher._batch_setitems..)r"_batch_setitemsiterr r)ritemsrrrr rMs  zHasher._batch_setitemscCst|t|dSr)r"r8r)rZ set_itemsrrr save_setszHasher.save_set)r)T)rrrrrrr8r>structr@rCr"dispatchcopyr3lenobjectr(r'rMrPsetrrrr r/s      rc@s"eZdZdZd ddZddZdS) NumpyHasherz1Special case the hasher for when numpy is loaded.rFcCs@||_tj||dddl}||_t|dr6|j|_nt|_dS)z Parameters ---------- hash_name: string The hash algorithm to be used coerce_mmap: boolean Make no difference between np.memmap and np.ndarray objects. r&rN getbuffer) coerce_mmaprrnumpynpr5rY _getbuffer memoryview)rr&rZr\rrr rs   zNumpyHasher.__init__cCst||jjr|jjs|jdkr*|}n&|jjr8|}n|jj rH|j }n|}|j | ||jj|jrt||jjr|jj}n|j}|d|j|j|jff}n6t||jjr|j dd|j t|dSt||dS)zSubclass the save method, to hash ndarray subclass, rather than pickling them. Off course, this is a total abuse of the Pickler class. rZHASHEDZ _HASHED_DTYPEzutf-8N)r0r\ZndarrayZdtypeZ hasobjectshapeflattenflags c_contiguous f_contiguousTr%r+r]viewZuint8rZZmemmapr7stridesencoder(r.rr8)rr-Zobj_c_contiguousklassrrr r8s$   zNumpyHasher.saveN)rF)rrrrrr8rrrr rWs rWrFcCsHd}||vrtd||dtjvr4t||d}n t|d}||S)aIQuick calculation of a hash to identify uniquely Python objects containing numpy arrays. Parameters ---------- hash_name: 'md5' or 'sha1' Hashing algorithm used. sha1 is supposedly safer, but md5 is faster. coerce_mmap: boolean Make no difference between np.memmap and np.ndarray )rsha1zAValid options for 'hash_name' are {}. Got hash_name={!r} instead.r[)r&rZrX) ValueErrorformatrErFrWrr)r-r&rZZvalid_hash_namesZhasherrrr rs   r)rF)rrr#rr(rQrEr1_Picklerr"rUrrrrWrrrrr s oY