U cN(@sVdZddlmZddlZddlmZmZmZmZm Z ddl Z ddl m Z ddlmZddlmZmZddlmZmZdd lmZmZmZmZmZerdd lmZmZmZm Z m!Z!d Z"d d ddddZ#dde"dfdddddddddZ$de"fdddddddZ%d dddd!d"d#Z&de"dfd$ddddd%d&d'Z'de"dfd(ddddd%d)d*Z(dS)+z" data hash pandas / numpy objects ) annotationsN) TYPE_CHECKINGHashableIterableIteratorcast)lib)hash_object_array) ArrayLikenpt)is_categorical_dtype is_list_like) ABCDataFrameABCExtensionArrayABCIndex ABCMultiIndex ABCSeries) Categorical DataFrameIndex MultiIndexSeriesZ0123456789123456zIterator[np.ndarray]intznpt.NDArray[np.uint64])arrays num_itemsreturncCsz t|}Wn$tk r0tjgtjdYSXt|g|}td}t|td}t|D]6\}}||}||N}||9}|td||7}qf|d|kst d|td7}|S)z Parameters ---------- arrays : Iterator[np.ndarray] num_items : int Returns ------- np.ndarray[uint64] Should be the same as CPython's tupleobject.c dtypeiCBixV4iXBzFed in wrong num_itemsi|) next StopIterationnparrayuint64 itertoolschainZ zeros_like enumerateAssertionError)rrfirstZmultoutiaZ inverse_ir,sz%hash_pandas_object..Nc3s"|]\}}t|jVqdSrB) hash_arrayr<)r>r?Zseries)r6r4r5r,r-rAsc3s$|]}tjddjVqdSr9r:r=r@r,r-rAsrcss|] }|VqdSrBr,)r>xr,r,r-rAszUnexpected type for hashing )pandasr_default_hash_key isinstancer hash_tuplesrrDr<astyperr$r%r.r3ritemslencolumns TypeErrortype)r2r3r4r5r6rhZserZ index_iterrhashesrZindex_hash_generator_hashesr,r@r-r;TsN           r;z+MultiIndex | Iterable[tuple[Hashable, ...]])valsr4r5rcszt|stdddlmm}t|ts6||n|fddtj D}fdd|D}t |t |}|S)a Hash an MultiIndex / listlike-of-tuples efficiently. Parameters ---------- vals : MultiIndex or listlike-of-tuples encoding : str, default 'utf8' hash_key : str, default _default_hash_key Returns ------- ndarray[np.uint64] of hashed values z'must be convertible to a list-of-tuplesr)rrcs(g|] }j|j|dddqS)FTZorderedZfastpath)codeslevels)r>level)rmir,r- szhash_tuples..c3s|]}t|dVqdS)r4r5N)_hash_categorical)r>catrZr,r-rAszhash_tuples..) r rNrFrrrHr from_tuplesrangeZnlevelsr.rL)rSr4r5rZcat_valsrQrPr,)rr4r5rXr-rIs    rIr)r\r4r5rcCsdt|jj}t|||dd}|}t|r<||j}ntj t|dd}| r`t j ||<|S)a Hash a Categorical by hashing its categories, and then mapping the codes to the hashes Parameters ---------- cat : Categorical encoding : str hash_key : str Returns ------- ndarray[np.uint64] of hashed values, same size as len(c) F)r6r#r) r!Zasarray categoriesr<rDZisnarLZtakerUzerosanyrZu8max)r\r4r5valueshashedmaskresultr,r,r-r[s  r[r )rSr4r5r6rcCs~t|dstd|j}t|r6td|}t|||St|trN|\}}n"t|t j sptdt |j dt ||||S)aK Given a 1d array, return an array of deterministic integers. Parameters ---------- vals : ndarray or ExtensionArray encoding : str, default 'utf8' Encoding for data & key when strings. hash_key : str, default _default_hash_key Hash_key for string key to encode. categorize : bool, default True Whether to first categorize object arrays before hashing. This is more efficient when the array contains duplicate values. Returns ------- ndarray[np.uint64, ndim=1] Hashed values, same length as the vals. rzmust pass a ndarray-likerz6hash_array requires np.ndarray or ExtensionArray, not z!. Use hash_pandas_object instead.)hasattrrNrr rr[rHrZ_values_for_factorizer!ZndarrayrO__name__ _hash_ndarray)rSr4r5r6rr?r,r,r-rDs     rDz np.ndarrayc Cst|j}t|tjr4tt|dtt|S|tkrH|d}nt |j tj tj frr| djddd}nt |j tjr|jdkr| d|jjd}n|rdd lm}m}m}||dd \}} |||| dd d } t| ||Szt|||}Wn.tk r.t|tt||}YnX||d ?N}|td9}||d?N}|td9}||d?N}|S)z! See hash_array.__doc__. u8i8Fr8ur)rr factorize)sortTrTle9z lb&&& )rr!Z issubdtypeZ complex128rDrealimagr0rJ issubclassrOZ datetime64Z timedelta64viewnumberitemsizerFrrrnZ _with_inferr[r rNr1objectr#) rSr4r5r6rrrrnrUr_r\r,r,r-rh7sB        rh))__doc__ __future__rr$typingrrrrrZnumpyr!Z pandas._libsrZpandas._libs.hashingr Zpandas._typingr r Zpandas.core.dtypes.commonr r Zpandas.core.dtypes.genericrrrrrrFrrrrrrGr.r;rIr[rDrhr,r,r,r-s<    $^.*3