B \G@sdZddlmZmZddlmZddlmZddlZ ddl m Z ddl m Z dd lmZd gZGd d d e eeZGd ddeZdd ZdddZdS)zUtilities for meta-estimators)ABCMetaabstractmethod) attrgetter)update_wrapperN) safe_indexing)six) BaseEstimatorif_delegate_has_methodcsJeZdZdZeddZd fdd ZfddZd d Zd d Z Z S)_BaseCompositionzOHandles parameter management for classifiers composed of named estimators. cCsdS)N)selfr r ;lib/python3.7/site-packages/sklearn/utils/metaestimators.py__init__sz_BaseComposition.__init__Tc s~tt|j|d}|s|St||}||xJ|D]B\}}t|dr4x.t|jddD]\}}||d||f<qZWq4W|S)N)deep get_paramsTz%s__%s)superr rgetattrupdatehasattrrZ iteritems) r attrroutZ estimatorsname estimatorkeyvalue) __class__r r _get_paramss   z_BaseComposition._get_paramsc s||krt||||t||}g}|r8t|\}}x:tt|D](}d|krH||krH|||||qHWtt |j f||S)N__) setattrpoprziplistrZiterkeys_replace_estimatorrr Z set_params)r rZparamsitemsnames_r)rr r _set_params&s  z_BaseComposition._set_paramscCsNtt||}x.t|D]"\}\}}||kr||f||<PqWt|||dS)N)r"r enumerater)r rrZnew_valZnew_estimatorsiZestimator_namer&r r rr#7s  z#_BaseComposition._replace_estimatorcCsvtt|t|kr&tdt|t||jdd}|rRtdt|dd|D}|rrtd|dS)Nz$Names provided are not unique: {0!r}F)rz:Estimator names conflict with constructor arguments: {0!r}cSsg|]}d|kr|qS)rr ).0rr r r Hsz4_BaseComposition._validate_names..z.Estimator names must not contain __: got {0!r})lenset ValueErrorformatr" intersectionrsorted)r r% invalid_namesr r r_validate_names@s  z _BaseComposition._validate_names)T) __name__ __module__ __qualname____doc__rrrr'r#r3 __classcell__r r )rrr s    r c@s"eZdZdZddZdddZdS)_IffHasAttrDescriptoratImplements a conditional property using the descriptor protocol. Using this class to create a decorator will raise an ``AttributeError`` if none of the delegates (specified in ``delegate_names``) is an attribute of the base object or the first found delegate does not have an attribute ``attribute_name``. This allows ducktyping of the decorated method based on ``delegate.attribute_name``. Here ``delegate`` is the first item in ``delegate_names`` for which ``hasattr(object, delegate) is True``. See https://docs.python.org/3/howto/descriptor.html for an explanation of descriptors. cCs ||_||_||_t||dS)N)fndelegate_namesattribute_namer)r r:r;r<r r rr]sz_IffHasAttrDescriptor.__init__Nc s~dk r`xVjD]:}yt|}Wntk r:wYqXt|jPqWtjdfdd}t|j|S)Ncsjf||S)N)r:)argskwargs)objr r rvsz/_IffHasAttrDescriptor.__get__..)r;rAttributeErrorrr<rr:)r r@typeZ delegate_namedelegaterr )r@r r__get__es   z_IffHasAttrDescriptor.__get__)N)r4r5r6r7rrEr r r rr9Nsr9cs.ttrttts"ffddS)aCreate a decorator for methods that are delegated to a sub-estimator This enables ducktyping by hasattr returning True according to the sub-estimator. Parameters ---------- delegate : string, list of strings or tuple of strings Name of the sub-estimator that can be accessed as an attribute of the base object. If a list or a tuple of names are provided, the first sub-estimator that is an attribute of the base object will be used. cst||jdS)N)r<)r9r4)r:)rDr rrAsz(if_delegate_has_method..) isinstancer"tuple)rDr )rDrr |s   cCst|ddrft|dstd|jd|jdkr:td|dkrT|t||}qp|t||}n t||}|dk rt||}nd}||fS) aCreate subset of dataset and properly handle kernels. Slice X, y according to indices for cross-validation, but take care of precomputed kernel-matrices or pairwise affinities / distances. If ``estimator._pairwise is True``, X needs to be square and we slice rows and columns. If ``train_indices`` is not None, we slice rows using ``indices`` (assumed the test set) and columns using ``train_indices``, indicating the training set. Labels y will always be indexed only along the first axis. Parameters ---------- estimator : object Estimator to determine whether we should slice only rows or rows and columns. X : array-like, sparse matrix or iterable Data to be indexed. If ``estimator._pairwise is True``, this needs to be a square array-like or sparse matrix. y : array-like, sparse matrix or iterable Targets to be indexed. indices : array of int Rows to select from X and y. If ``estimator._pairwise is True`` and ``train_indices is None`` then ``indices`` will also be used to slice columns. train_indices : array of int or None, default=None If ``estimator._pairwise is True`` and ``train_indices is not None``, then ``train_indices`` will be use to slice the columns of X. Returns ------- X_subset : array-like, sparse matrix or list Indexed data. y_subset : array-like, sparse matrix or list Indexed targets. Z _pairwiseFshapezXPrecomputed kernels or affinity matrices have to be passed as arrays or sparse matrices.rz"X should be a square kernel matrixN)rrr.rHnpZix_r)rXyindicesZ train_indicesZX_subsetZy_subsetr r r _safe_splits,    rN)N)r7abcrroperatorr functoolsrZnumpyrJZutilsrZ externalsrbaser __all__Zwith_metaclassr objectr9r rNr r r rs     <.