codekingpro/portable-devtools
114k
1�
2 z�j$� � � � S r SSKrSSKrSSKJr SSKJs Jr SSK J3r4 SSK J r / SQr
S r\5" \5 S 5 rS rS rS6 rS rS7S jrS7S
jrS8S jrS rS rS9S jrS:S jrS;S jr S<S jr \7" \ 5 S=S j5 r S8S jr \8" \ 5 S:S j5 r S:S jr S r!\9" \!5 S 5 r"S r#\10" \#5 S 5 r$ S<S jr%\11" \%5 S>S j5 r&S;S jr'\12" \'5 S;S j5 r(S8S! jr)\13" \)5 S?S" j5 r*S@S# jr+S$ r, SAS% jr-\14" \-5 SBS& j5 r. SCS' jr/\15" \/5 SDS( j5 r0S) r1\16" \15 S* 5 r2S;S+ jr3\17" \35 SES, j5 r4S- r5\18" \55 S. 5 r6 S<S/ jr7\19" \75 SFS0 j5 r8 SAS1 jr9\20" \95 SGS2 j5 r: SHS3 jr;\21" \;5 SIS4 j5 r< S<S5 jr=\22" \=5 SJS6 j5 r>C23g)Kz�24Collection of utilities to manipulate structured arrays.25 26Most of these functions were initially implemented by John Hunter for27matplotlib. They have been rewritten and extended for convenience.28 29� N)�array_function_dispatch)�_is_string_like)�
append_fields�apply_along_fields�assign_fields_by_name� drop_fields�find_duplicates�
flatten_descr�get_fieldstructure� get_names�get_names_flat�join_by� merge_arrays�rec_append_fields�rec_drop_fields�rec_join�recursive_fill_fields�
rename_fields�
repack_fields�require_fields� stack_arrays�structured_to_unstructured�unstructured_to_structuredc � � X4$ �N� )�input�outputs �\D:\code\apps\devtools\python\user_packages\Python313\site-packages\numpy/lib/recfunctions.py�!_recursive_fill_fields_dispatcherr s30 � � �?�� c �� � UR nUR H? n X nUR R b [ XAU 5 M/ XAU S[ U5 & MA U$ ! [ a MR f = f)a5 31Fills fields from output with fields from input,32with support for nested structures.33 34Parameters35----------36input : ndarray37 Input array.38output : ndarray39 Output array.40 41Notes42-----43* `output` should be at least the same size as `input`44 45Examples46--------47>>> import numpy as np48>>> from numpy.lib import recfunctions as rfn49>>> a = np.array([(1, 10.), (2, 20.)], dtype=[('A', np.int64), ('B', np.float64)])50>>> b = np.zeros((3,), dtype=a.dtype)51>>> rfn.recursive_fill_fields(a, b)52array([(1, 10.), (2, 20.), (0, 0.)], dtype=[('A', '<i8'), ('B', '<f8')])53 54N)�dtype�names�55ValueErrorr �len)r r �newdtype�field�currents r r r sr � �6 �|�|�H�� � �� ��l�G� �=�=� � � *� !�'�%�=� 9�+2�5�M�-�3�w�<� (� � �M��
� � � �s �A � 56A,�+A,c �� ^ � T R c ST 4/$ U 4S jT R 5 nU VVs/ s H" u p#[ U5 S:X a UOUS U4US 4PM$ snn$ s snnf )a5 57Produce a list of name/dtype pairs corresponding to the dtype fields58 59Similar to dtype.descr, but the second item of each tuple is a dtype, not a60string. As a result, this handles subarray dtypes61 62Can be passed to the dtype constructor to reconstruct the dtype, noting that63this (deliberately) discards field offsets.64 65Examples66--------67>>> import numpy as np68>>> dt = np.dtype([(('a', 'A'), np.int64), ('b', np.double, 3)])69>>> dt.descr70[(('a', 'A'), '<i8'), ('b', '<f8', (3,))]71>>> _get_fieldspec(dt)72[(('a', 'A'), dtype('int64')), ('b', dtype(('<f8', (3,))))]73 74� c 3 �D ># � U H oTR U 4v � M g 7fr )�fields)�.0�namer# s �r � <genexpr>�!_get_fieldspec.<locals>.<genexpr>_ s �� � �E���� � �d�+�,��s � � r )r$ r& )r# r- r/ �fs ` r �_get_fieldspecr4 G so �� �(
�{�{���U� �}� �E����E�� "�75�!��� ��V�q�[�T�q��t�T�l�A�a�D� 9�!�76� 77�� 78s �)A c �� � / nU R nU HL nX nUR b' UR U[ [ U5 5 45 M; UR U5 MN [ U5 $ )a 79Returns the field names of the input datatype as a tuple. Input datatype80must have fields otherwise error is raised.81 82Parameters83----------84adtype : dtype85 Input datatype86 87Examples88--------89>>> import numpy as np90>>> from numpy.lib import recfunctions as rfn91>>> rfn.get_names(np.empty((1,), dtype=[('A', int)]).dtype)92('A',)93>>> rfn.get_names(np.empty((1,), dtype=[('A',int), ('B', float)]).dtype)94('A', 'B')95>>> adtype = np.dtype([('a', int), ('b', [('ba', int), ('bb', int)])])96>>> rfn.get_names(adtype)97('a', ('b', ('ba', 'bb')))98)r$ �append�tupler ��adtype� listnamesr$ r/ r) s r r r g se � �, �I� �L�L�E����,�� �=�=� $� � � �d�E�)�G�*<�$=� >� ?� � � �T� "� � �� �r! c �� � / nU R nU HA nUR U5 X nUR c M' UR [ U5 5 MC [ U5 $ )aU 99Returns the field names of the input datatype as a tuple. Input datatype100must have fields otherwise error is raised.101Nested structure are flattened beforehand.102 103Parameters104----------105adtype : dtype106 Input datatype107 108Examples109--------110>>> import numpy as np111>>> from numpy.lib import recfunctions as rfn112>>> rfn.get_names_flat(np.empty((1,), dtype=[('A', int)]).dtype) is None113False114>>> rfn.get_names_flat(np.empty((1,), dtype=[('A',int), ('B', str)]).dtype)115('A', 'B')116>>> adtype = np.dtype([('a', int), ('b', [('ba', int), ('bb', int)])])117>>> rfn.get_names_flat(adtype)118('a', 'b', 'ba', 'bb')119)r$ r6 �extendr
r7 r8 s r r
r
� s[ � �. �I� �L�L�E������� ��,�� �=�=� $� � � �^�G� 4� 5� �120 �� �r! c �� � U R nUc SU 44$ / nU HO nU R U u pEUR b UR [ U5 5 M= UR X445 MQ [ U5 $ )a; 121Flatten a structured data-type description.122 123Examples124--------125>>> import numpy as np126>>> from numpy.lib import recfunctions as rfn127>>> ndtype = np.dtype([('a', '<i4'), ('b', [('ba', '<f8'), ('bb', '<i4')])])128>>> rfn.flatten_descr(ndtype)129(('a', dtype('int32')), ('ba', dtype('float64')), ('bb', dtype('int32')))130 131r+ )r$ r- r< r132 r6 r7 )�ndtyper$ �descrr( �typ�_s r r133 r134 � sr � �
�L�L�E� �}��V� �� ����E� �}�}�U�+�H�S��y�y�$�� � �]�3� /�0�� � �e�\�*� � �U�|�r! c �p � / nU( a. U H' nUR [ UR 5 5 M) OjU Hd nUR nUR b5 [ UR 5 S:X a UR [ U5 5 MQ UR
SU45 Mf [ R " U5 $ )N� r+ )r< r135 r# r$ r& r4 r6 �np)� seqarrays�flattenr' �ar) s r �136_zip_dtyperH � s� � ��H���A� �O�O�M�!�'�'� 2� 3� � �A��g�g�G��}�}�(�S�� � �-?�1�-D�����w� 7�8�����W�
�.�
� �8�8�H� � r! c �( � [ XS9R $ )z�137Combine the dtype description of a series of arrays.138 139Parameters140----------141seqarrays : sequence of arrays142 Sequence of arrays143flatten : {boolean}, optional144 Whether to collapse nested descriptions.145�rF )rH r? )rE rF s r �146_zip_descrrK � s � � �i� 1� 7� 7�7r! c �` � Uc 0 nU R nU H� nX nUR b. U( a U/X$' O/ X$' UR [ XTU5 5 MB [ UR U/ 5 =( d / 5 nU( a UR U5 O147U( a U/nU=( d / X$' M� U$ )a) 148Returns a dictionary with fields indexing lists of their parent fields.149 150This function is used to simplify access to fields nested in other fields.151 152Parameters153----------154adtype : np.dtype155 Input datatype156lastname : optional157 Last processed field name (used internally during recursion).158parents : dictionary159 Dictionary of parent fields (used internally during recursion).160 161Examples162--------163>>> import numpy as np164>>> from numpy.lib import recfunctions as rfn165>>> ndtype = np.dtype([('A', int),166... ('B', [('BA', int),167... ('BB', [('BBA', int), ('BBB', int)])])])168>>> rfn.get_fieldstructure(ndtype)169... # XXX: possible regression, order of BBA and BBB is swapped170{'A': [], 'B': [], 'BA': ['B'], 'BB': ['B'], 'BBA': ['B', 'BB'], 'BBB': ['B', 'BB']}171 172)r$ �updater �list�getr6 )r9 �lastname�parentsr$ r/ r) �173lastparents r r r � s� � �6 ���� �L�L�E����,�� �=�=� $��!)� ��
� "��
� �N�N�-�g�W�E� F� �g�k�k�(�B� 7� =�2�>�J���!�!�(�+�� &�\�174� &� ,�"�G�M� � �Nr! c # � # � U HD n[ U[ R 5 ( a [ [ U5 5 Sh v�N M@ Uv � MF g N7f)zi175Returns an iterator of concatenated fields from a sequence of arrays,176collapsing any nested structure.177 178N)�179isinstancerD �void�_izip_fields_flatr7 ��iterable�elements r rV rV s< � � � �� �g�r�w�w� '� '�(��w� �8� 8� 8��M� � 8�s �<A�A�Ac # �< # � U H� n[ US5 ( a* [ U[ 5 ( d [ U5 Sh v�N M> [ U[ R180 5 ( a- [
[ U5 5 S:X a [ U5 Sh v�N M� Uv � M� g N[ N7f)zH181Returns an iterator of concatenated fields from a sequence of arrays.182 183�__iter__NrC )�hasattrrT �str� _izip_fieldsrD rU r& r7 rW s r r^ r^ sw � � �184 �� �G�Z� (� (� �w��,�,�#�G�,� ,� ,�
�����
)�
)�c�%��.�.A�Q�.F�#�G�,� ,� ,��M� �
-�
-�s" �:B �B�A B �B� B �B c # � # � U( a [ nO[ n[ R " U SU06 H n[ U" U5 5 v � M g7f)a 185Returns an iterator of concatenated items from a sequence of arrays.186 187Parameters188----------189seqarrays : sequence of arrays190 Sequence of arrays.191fill_value : {None, integer}192 Value used to pad shorter iterables.193flatten : {True, False},194 Whether to195� fillvalueN)rV r^ � itertools� zip_longestr7 )rE �196fill_valuerF �zipfunc�tups r �
_izip_recordsrf / s> � � � �#�� ���$�$�i�F�:�F���G�C�L�!�!� G�s �AAc � � [ U [ R 5 ( d SnU( a( U( a U R [ R197 5 n U $ [ R " U 5 n U( a U R [ R 5 n U $ )zt198Private function: return a recarray, a ndarray, a MaskedArray199or a MaskedRecords depending on the input parameters200F) rT �ma� MaskedArray�view�mrec�
MaskedRecords�filledrD �recarray)r �usemask�201asrecarrays r � _fix_outputrq G se � �202 �f�b�n�n� -� -���� ��[�[��!3�!3�4�F�203 �M� ���6�"�� ��[�[����-�F� �Mr! c �� � U R R nU R U R U R pTnU=( d 0 R 5 H u pgXb; d M XuU' XsU XF ' M U $ )zd204Update the fill_value and masked data of `output`205from the default given in a dictionary defaults.206)r# r$ �data�maskrc �items)r �defaultsr$ rs rt rc �k�vs r �
_fix_defaultsry X sf � �207
�L�L� � �E� &� � �V�[�[�&�:K�:K��T��>�r�(�(�*��� �:� �q�M� ��G�D�G� � +� �Mr! c � � U $ r r )rE rc rF ro rp s r �_merge_arrays_dispatcherr{ f s � � �r! c 208 �l � [ U 5 S:X a [ R " U S 5 n [ U [ R [ R209 45 ( a� U R nUR c [ R " SU4/5 nU( a [ U 4SS9U:X aw U R 5 n U( a) U( a [ R nO9[ R nO(U( a [ R nO[ R nU R XVS9$ U 4n O&U Vs/ s H n[ R " U5 PM n n[! S U 5 5 n[# U5 n [ XS9n210/ n / n U( Ga� [% X5 GHi u p�X�-211 nU
R 5 R' 5 n[ R( " U
5 R 5 nU( a� [ R* " X R 5 n[ U[ R [ R212 45 ( an [ UR 5 S:X a UR- 5 S nSnOD[ R. " UU
R SS 9n[ R0 " S213UR S 9nOSnSnU R3 [4 R6 " UU/U- 5 5 U R3 [4 R6 " UW/U- 5 5 GMl [! [9 X�S95 n[ R. " [ R: " UX�S 9[= [9 X�S95 S
9nU( a UR [ R 5 nU$ [% X5 H� u p�X�-214 nU
R 5 R' 5 nU( a� [ R* " X R 5 n[ U[ R [ R215 45 ( aM [ UR 5 S:X a UR- 5 S nO#[ R. " UU
R SS 9nOSnU R3 [4 R6 " UU/U- 5 5 M� [ R: " [! [9 X�S95 X�S 9nU( a UR [ R 5 nU$ s snf )a� 216Merge arrays field by field.217 218Parameters219----------220seqarrays : sequence of ndarrays221 Sequence of arrays222fill_value : {float}, optional223 Filling value used to pad missing data on the shorter arrays.224flatten : {False, True}, optional225 Whether to collapse nested fields.226usemask : {False, True}, optional227 Whether to return a masked array or not.228asrecarray : {False, True}, optional229 Whether to return a recarray (MaskedRecords) or not.230 231Examples232--------233>>> import numpy as np234>>> from numpy.lib import recfunctions as rfn235>>> rfn.merge_arrays((np.array([1, 2]), np.array([10., 20., 30.])))236array([( 1, 10.), ( 2, 20.), (-1, 30.)],237 dtype=[('f0', '<i8'), ('f1', '<f8')])238 239>>> rfn.merge_arrays((np.array([1, 2], dtype=np.int64),240... np.array([10., 20., 30.])), usemask=False)241 array([(1, 10.0), (2, 20.0), (-1, 30.0)],242 dtype=[('f0', '<i8'), ('f1', '<f8')])243>>> rfn.merge_arrays((np.array([1, 2]).view([('a', np.int64)]),244... np.array([10., 20., 30.])),245... usemask=False, asrecarray=True)246rec.array([( 1, 10.), ( 2, 20.), (-1, 30.)],247 dtype=[('a', '<i8'), ('f1', '<f8')])248 249Notes250-----251* Without a mask, the missing value will be filled with something,252 depending on what its corresponding type:253 254 * ``-1`` for integers255 * ``-1.0`` for floating point numbers256 * ``'-'`` for characters257 * ``'-1'`` for strings258 * ``True`` for boolean values259* XXX: I just obtained these values empirically260rC r Nr+ TrJ )r# �typec 3 �8 # � U H oR v � M g 7fr )�size)r. rG s r r0 � merge_arrays.<locals>.<genexpr>� s � � �,�)�Q�&�&�)�s �)r# �ndmin)rC �r# )r# �count)rt ) r& rD �261asanyarrayrT �ndarrayrU r# r$ rH �ravelrk rl rh ri rn rj r7 �max�zip� __array__� getmaskarray�_check_fill_value�item�array�onesr6 ra �chainrf �fromiterrN )rE rc rF ro rp �seqdtype�seqtype�_m�sizes� maxlengthr' �seqdata�seqmaskrG �n� nbmissingrs rt �fval�fmskr s r r r k sr � �d �I��!���M�M�)�A�,�/� ��)�b�j�j�"�'�'� 2�3�3� �?�?�� �>�>� !��x�x�"�h� � 0�1�H��*�i�\�4�@�H�L�!���)�I�� � "� 0� 0�G� �n�n�G�� �+�+�� �*�*�� �>�>��>�?� ?�"� �I� 2;�;��2�R�]�]�2�&�� �;� �,�)�,� ,�E��E�262�I��)�5�H��G��G���)�+�F�Q�"� �I��7�7�9�&�&�(�D��?�?�1�%�+�+�-�D���-�-�j�'�'�B�� �d�R�Z�Z����$9�:�:��4�:�:��!�+� #�y�y�{�1�~�� #�� !�x�x��A�G�G�1� E�� !�w�w�t�4�:�:� >������� �N�N�9�?�?�4�$��)�1C�D� E� �N�N�9�?�?�4�$��)�1C�D� E�) ,�, �]�7�<�=�����"�+�+�d�(�L� #�M�'�$K� L�N�� ��[�[��!3�!3�4�F�* �M�% �)�+�F�Q�"� �I��7�7�9�&�&�(�D���-�-�j�'�'�B�� �d�R�Z�Z����$9�:�:��4�:�:��!�+� #�y�y�{�1�~�� !�x�x��A�G�G�1� E����� �N�N�9�?�?�4�$��)�1C�D� E� ,� ���U�=��#J� K�#+�>�� ��[�[����-�F� �M��u <s � R1c � � U 4$ r r )�base�263drop_namesro rp s r �_drop_fields_dispatcherr� � � � � �7�Nr! c �� ^� [ U5 ( a U/nO [ U5 nU4S jmT" U R U5 n[ R " U R264 US9n[
X5 n[ XRUS9$ )a 265Return a new array with fields in `drop_names` dropped.266 267Nested fields are supported.268 269Parameters270----------271base : array272 Input array273drop_names : string or sequence274 String or sequence of strings corresponding to the names of the275 fields to drop.276usemask : {False, True}, optional277 Whether to return a masked array or not.278asrecarray : string or sequence, optional279 Whether to return a recarray or a mrecarray (`asrecarray=True`) or280 a plain ndarray or masked array with flexible dtype. The default281 is False.282 283Examples284--------285>>> import numpy as np286>>> from numpy.lib import recfunctions as rfn287>>> a = np.array([(1, (2, 3.0)), (4, (5, 6.0))],288... dtype=[('a', np.int64), ('b', [('ba', np.double), ('bb', np.int64)])])289>>> rfn.drop_fields(a, 'a')290array([((2., 3),), ((5., 6),)],291 dtype=[('b', [('ba', '<f8'), ('bb', '<i8')])])292>>> rfn.drop_fields(a, 'ba')293array([(1, (3,)), (4, (6,))], dtype=[('a', '<i8'), ('b', [('bb', '<i8')])])294>>> rfn.drop_fields(a, ['ba', 'bb'])295array([(1,), (4,)], dtype=[('a', '<i8')])296c �� >� U R n/ nU HR nX nXA; a M UR b% T" XQ5 nU( a UR XF45 M> M@ UR XE45 MT U$ r )r$ r6 )r> r� r$ r' r/ r) r? � _drop_descrs �r r� � drop_fields.<locals>._drop_descr sl �� �� � �����D� �l�G��!���}�}�(�#�G�8��� �O�O�T�M�2� � �����0� � �r! r� �ro rp )r �setr# rD �empty�shaper rq )r� r� ro rp r' r r� s @r r r � sc �� �F �z�"�"� �\�297���_�298�
� �4�:�:�z�2�H�
�X�X�d�j�j��
1�F�
"�4�
0�F� �v�:� F�Fr! c � � U Vs/ s H oDU R U 4PM nn[ R " U R US9n[ X5 n[ XbUS9$ s snf )aL 299Return a new array keeping only the fields in `keep_names`,300and preserving the order of those fields.301 302Parameters303----------304base : array305 Input array306keep_names : string or sequence307 String or sequence of strings corresponding to the names of the308 fields to keep. Order of the names will be preserved.309usemask : {False, True}, optional310 Whether to return a masked array or not.311asrecarray : string or sequence, optional312 Whether to return a recarray or a mrecarray (`asrecarray=True`) or313 a plain ndarray or masked array with flexible dtype. The default314 is False.315r� r� )r# rD r� r� r rq )r� �316keep_namesro rp r� r' r s r � _keep_fieldsr� 6 sS � �& -7�7�J�q�D�J�J�q�M�"�J�H�7�
�X�X�d�j�j��
1�F�
"�4�
0�F� �v�:� F�F�� 8s �Ac � � U 4$ r r �r� r� s r �_rec_drop_fields_dispatcherr� O r� r! c � � [ XSSS9$ )zC317Returns a new numpy.recarray with fields in `drop_names` dropped.318FTr� )r r� s r r r S s � �319 �t��4� H�Hr! c � � U 4$ r r )r� �320namemappers r �_rename_fields_dispatcherr� [ r� r! c �X ^� U4S jmT" U R U5 nU R U5 $ )a� 321Rename the fields from a flexible-datatype ndarray or recarray.322 323Nested fields are supported.324 325Parameters326----------327base : ndarray328 Input array whose fields must be modified.329namemapper : dictionary330 Dictionary mapping old field names to their new version.331 332Examples333--------334>>> import numpy as np335>>> from numpy.lib import recfunctions as rfn336>>> a = np.array([(1, (2, [3.0, 30.])), (4, (5, [6.0, 60.]))],337... dtype=[('a', int),('b', [('ba', float), ('bb', (float, 2))])])338>>> rfn.rename_fields(a, {'a':'A', 'bb':'BB'})339array([(1, (2., [ 3., 30.])), (4, (5., [ 6., 60.]))],340 dtype=[('A', '<i8'), ('b', [('ba', '<f8'), ('BB', '<f8', (2,))])])341 342c �� >� / nU R HR nUR X35 nX nUR b UR UT" XQ5 45 M@ UR XE45 MT U$ r )r$ rO r6 )r> r� r' r/ �newnamer) �_recursive_rename_fieldss �r r� �/rename_fields.<locals>._recursive_rename_fieldsx sg �� ����L�L�D� �n�n�T�0�G� �l�G��}�}�(���� � 6�w� K�L�� ���� 2�3� !� �r! )r# rj )r� r� r' r� s @r r r _ s) �� �2 � (��343�344�J�?�H� �9�9�X� � r! c # �, # � U v � U S h v�N g N7fr r )r� r$ rs �dtypesrc ro rp s r �_append_fields_dispatcherr� � s � � �345�J��O�O�� �346��c �Z � [ U[ [ 45 ( a& [ U5 [ U5 :w a
Sn[ U5 eO[ U[347 5 ( a U/nU/nUca U Vs/ s H n[ R " USSS9PM nn[ X5 V Vs/ s H" u p�UR X�R 4/5 PM$ nn nO�[ U[ [ 45 ( d U/n[ U5 [ U5 :w a+ [ U5 S:X a U[ U5 - nO
Sn[ U5 e[ X!U5 VV348V s/ s H. u p�n [ R " USSU S9R X�4/5 PM0 nn349nn [ XUS9n [ U5 S:� a
[ USUUS 9nOUR 5 n[ R " [ [ U 5 [ U5 5 [! U R 5 [! UR 5 - S3509n [# X 5 n [# X,5 n [% X�US 9$ s snf s snn f s sn n351nf ) a� 352Add new fields to an existing array.353 354The names of the fields are given with the `names` arguments,355the corresponding values with the `data` arguments.356If a single field is appended, `names`, `data` and `dtypes` do not have357to be lists but just values.358 359Parameters360----------361base : array362 Input array to extend.363names : string, sequence364 String or sequence of strings corresponding to the names365 of the new fields.366data : array or sequence of arrays367 Array or sequence of arrays storing the fields to add to the base.368dtypes : sequence of datatypes, optional369 Datatype or sequence of datatypes.370 If None, the datatypes are estimated from the `data`.371fill_value : {float}, optional372 Filling value used to pad missing data on the shorter arrays.373usemask : {False, True}, optional374 Whether to return a masked array or not.375asrecarray : {False, True}, optional376 Whether to return a recarray (MaskedRecords) or not.377 378z7The number of arrays does not match the number of namesNT)�copy�subokrC z5The dtypes argument must be None, a dtype, or a list.)r� r� r# )ro rc )rF ro rc r� r� )rT r7 rN r& r% r] rD r� r� rj r# r �poprh �379masked_allr� r4 r rq )
r� r$ rs r� rc ro rp �msgrG r/ r� �dr s
r r r � s� � �@ �%�%�� �'�'� �u�:��T�� "�K�C� �S�/� !� #� 380�E�3� � �� ���x��
�~�<@�A�D�q������T�2�D��A�;>�u�;K�L�;K�i�t�����w�w��(�)�;K��L���&�5�$�-�0�0� �Z�F� �t�9��F� � #��6�{�a� � �#�d�)�+��M�� ��o�%�!$�T�&�!9�;�!9�I�Q�1� �����T��;�@�@�1�&��J�!9�
� ;� ��*� E�D�381�4�y�1�}��D�$��'1�3�� �x�x�z��
�]�]� �C��I�s�4�y�!� �T�Z�Z�(�>�$�*�*�+E�E�G�F� #�4�
0�F�
"�4�
0�F� �v�:� F�F��7 B��L��;s �$ H�)H �25H&c # �, # � U v � U S h v�N g N7fr r �r� r$ rs r� s r � _rec_append_fields_dispatcherr� � s � � �382�J��O�O�r� c � � [ XX#SSS9$ )a� 383Add new fields to an existing array.384 385The names of the fields are given with the `names` arguments,386the corresponding values with the `data` arguments.387If a single field is appended, `names`, `data` and `dtypes` do not have388to be lists but just values.389 390Parameters391----------392base : array393 Input array to extend.394names : string, sequence395 String or sequence of strings corresponding to the names396 of the new fields.397data : array or sequence of arrays398 Array or sequence of arrays storing the fields to add to the base.399dtypes : sequence of datatypes, optional400 Datatype or sequence of datatypes.401 If None, the datatypes are estimated from the `data`.402 403See Also404--------405append_fields406 407Returns408-------409appended_array : np.recarray410TF)rs r� rp ro )r r� s r r r � s � �> ��4�$(�%� 9� 9r! c � � U 4$ r r )rG �align�recurses r �_repack_fields_dispatcherr� � s � �
�4�Kr! c �� � [ U [ R 5 ( d$ [ U R XS9nU R USS9$ U R411 c U $ / nU R412 HU nU R U nU( a [ US USS9nOUS n[ U5 S:X a US U4nUR XW45 MW [ R " XAS9n[ R " U R U45 $ ) ac 413Re-pack the fields of a structured array or dtype in memory.414 415The memory layout of structured datatypes allows fields at arbitrary416byte offsets. This means the fields can be separated by padding bytes,417their offsets can be non-monotonically increasing, and they can overlap.418 419This method removes any overlaps and reorders the fields in memory so they420have increasing byte offsets, and adds or removes padding bytes depending421on the `align` option, which behaves like the `align` option to422`numpy.dtype`.423 424If `align=False`, this method produces a "packed" memory layout in which425each field starts at the byte the previous field ended, and any padding426bytes are removed.427 428If `align=True`, this methods produces an "aligned" memory layout in which429each field's offset is a multiple of its alignment, and the total itemsize430is a multiple of the largest alignment, by adding padding bytes as needed.431 432Parameters433----------434a : ndarray or dtype435 array or dtype for which to repack the fields.436align : boolean437 If true, use an "aligned" memory layout, otherwise use a "packed" layout.438recurse : boolean439 If True, also repack nested structures.440 441Returns442-------443repacked : ndarray or dtype444 Copy of `a` with fields repacked, or `a` itself if no repacking was445 needed.446 447Examples448--------449>>> import numpy as np450 451>>> from numpy.lib import recfunctions as rfn452>>> def print_offsets(d):453... print("offsets:", [d.fields[name][1] for name in d.names])454... print("itemsize:", d.itemsize)455...456>>> dt = np.dtype('u1, <i8, <f8', align=True)457>>> dt458dtype({'names': ['f0', 'f1', 'f2'], 'formats': ['u1', '<i8', '<f8'], 'offsets': [0, 8, 16], 'itemsize': 24}, align=True)459>>> print_offsets(dt)460offsets: [0, 8, 16]461itemsize: 24462>>> packed_dt = rfn.repack_fields(dt)463>>> packed_dt464dtype([('f0', 'u1'), ('f1', '<i8'), ('f2', '<f8')])465>>> print_offsets(packed_dt)466offsets: [0, 1, 9]467itemsize: 17468 469)r� r� F)r� r T� r2 �r� )470rT rD r# r �astyper$ r- r& r6 r} )rG r� r� �dt� fieldinfor/ re �fmts r r r s� � �z �a���� "� "�
�1�7�7�%�
A���x�x���x�'�'��w�w�����I������h�h�t�n�� � ��A��e�T�B�C��a�&�C� �s�8�q�=���F�D�>�D����$��%� � 471 ���)� )�B�
�8�8�Q�V�V�R�L� !�!r! c �� � S n/ nU R H� nU R U nUS US pvU" U5 u phUR c. UR [ R " Xh445 X�U- 45 M` [ XgU- 5 n UR n472[ U5 HK n U S:X a UR U 5 M UR U V V
Vs/ s H u p�o�X�X�- - 4PM snn
n 5 MM M� U$ s snn
n f )z�473Returns a flat list of (dtype, count, offset) tuples of all the474scalar fields in the dtype "dt", including nested fields, in left475to right order.476c � � SnU R S:w a5 U R H nX- nM U R n U R S:w a M5 X4$ )NrC r )r� r� )r� r� r s r �477count_elem�+_get_fields_and_offsets.<locals>.count_elem^ sG � ����h�h�"�n������
�� !����B� �h�h�"�n� �y�r! r rC ) r$ r- r6 rD r# �_get_fields_and_offsets�itemsize�ranger< )r� �offsetr� r- r/ r( �f_dt�f_offsetr� � subfieldsr �ir� �c�os r r� r� U s� � �� �F������ � �$��� �q��5��8�h� �T�"��� �:�:� � �M�M�2�8�8�T�4�L�1�1��6G�H� I�/���6G�H�I��=�=�D��1�X����6��M�M�)�,��M�M�y�"Q�y�G�A�!�q�a�h�,�#7�y�"Q�R� � �" �M�� #Rs �C,c �L � [ U 5 S:: a U$ U S U S : nU( a [ [ U 5 [ U5 5 nO [ X5 nSnSnU HB u pxUS:w a U( a gUc UnXb:w a gXxS-478 U- - n OUn Ub Xu-479 n480Uc U481nXj:w a gU nMD U( a U* $ U$ )z�482Returns the stride between the fields, or None if the stride is not483constant. The values in "counts" designate the lengths of484subarrays. Subarrays are treated as many contiguous fields, with485always positive stride.486rC r N)r& r� �reversed) �offsets�countsr� �negative�it� prev_offset�strider� r� �487end_offset�488new_strides r �_common_strider� z s� � � �7�|�q�����q�z�G�A�J�&�H��
��'�"�H�V�$4�
5��
��
!���K�
�F��
�� �A�:����~�!���!�� �1�9��"8�8�J� �J� � "� �-�J��~�#���#�� � �' �* ��w�� �Mr! c � � U 4$ r r )�arrr# r� �castings r �&_structured_to_unstructured_dispatcherr� � � � � �6�Mr! c �$ ^� U R R c [ S5 e[ U R 5 n[ U5 nUS:X a Uc [ S5 eUS:X a [ S5 e[
U6 u pgn[ U5 V s/ s H n SU 3PM489 n490n Uc. [ R " U V s/ s H o�R PM sn 6 mO[ R " U5 m[ R " U491UUU R R S.5 n U R U 5 n [ U 5 [ R [ R [ R 4; n
U( GdD U
( Ga< [# U4S jU 5 5 ( Ga! [% X�TR 5 nUGb U R& nU R( [+ U5 TR 4- nU R, [/ U5 S 4- nU S492[ R0 4 R [ R2 5 n U S493[5 U5 S24 n [ R6 R8 R; U UUS S 9n U R T5 S
n US: a494 U S495SSS24 n [ U 5 [ UR< 5 La U" U 5 n U $ [ R " U496U V s/ s H n TU R( 4PM sn S.5 nU R? UX#S9n U R T[+ U5 445 $ s sn f s sn f s sn f )a� 497Converts an n-D structured array into an (n+1)-D unstructured array.498 499The new array will have a new last dimension equal in size to the500number of field-elements of the input array. If not supplied, the output501datatype is determined from the numpy type promotion rules applied to all502the field datatypes.503 504Nested fields, as well as each element of any subarray fields, all count505as a single field-elements.506 507Parameters508----------509arr : ndarray510 Structured array or dtype to convert. Cannot contain object datatype.511dtype : dtype, optional512 The dtype of the output unstructured array.513copy : bool, optional514 If true, always return a copy. If false, a view is returned if515 possible, such as when the `dtype` and strides of the fields are516 suitable and the array subtype is one of `numpy.ndarray`,517 `numpy.recarray` or `numpy.memmap`.518 519 .. versionchanged:: 1.25.0520 A view can now be returned if the fields are separated by a521 uniform stride.522 523casting : {'no', 'equiv', 'safe', 'same_kind', 'unsafe'}, optional524 See casting argument of `numpy.ndarray.astype`. Controls what kind of525 data casting may occur.526 527Returns528-------529unstructured : ndarray530 Unstructured array with one more dimension.531 532Examples533--------534>>> import numpy as np535 536>>> from numpy.lib import recfunctions as rfn537>>> a = np.zeros(4, dtype=[('a', 'i4'), ('b', 'f4,u2'), ('c', 'f4', 2)])538>>> a539array([(0, (0., 0), [0., 0.]), (0, (0., 0), [0., 0.]),540 (0, (0., 0), [0., 0.]), (0, (0., 0), [0., 0.])],541 dtype=[('a', '<i4'), ('b', [('f0', '<f4'), ('f1', '<u2')]), ('c', '<f4', (2,))])542>>> rfn.structured_to_unstructured(a)543array([[0., 0., 0., 0., 0.],544 [0., 0., 0., 0., 0.],545 [0., 0., 0., 0., 0.],546 [0., 0., 0., 0., 0.]])547 548>>> b = np.array([(1, 2, 5), (4, 5, 7), (7, 8 ,11), (10, 11, 12)],549... dtype=[('x', 'i4'), ('y', 'f4'), ('z', 'f8')])550>>> np.mean(rfn.structured_to_unstructured(b[['x', 'z']]), axis=-1)551array([ 3. , 5.5, 9. , 11. ])552 553N� arr must be a structured arrayr z(arr has no fields. Unable to guess dtypez#arr with no fields is not supportedr3 �r$ �formatsr� r� c 3 �@ ># � U H oR T:H v � M g 7fr )r� )r. r� � out_dtypes �r r0 �-structured_to_unstructured.<locals>.<genexpr> s �� � �&J�c��w�w�)�';�c�s � rC .T)r� �.r ������r$ r� �r� r� ) r# r$ r% r� r& �NotImplementedErrorr� r� rD � result_typer� r� rj r} r� rn �memmap�allr� �__array_wrap__r� �sum�strides�abs�newaxis�uint8�min�lib�
stride_tricks�554as_strided�__self__r� )r� r# r� r� r- �n_fields�dtsr� r� r� r$ r� �flattened_fields�can_view�
common_stride�wrap� new_shape� new_strides�
packed_fieldsr� s @r r r � s� �� �x �y�y��� ��9�:�:�
$�S�Y�Y�
/�F��6�{�H��1�}�� ��C�D�D� �Q� �!�"G�H�H� ��<��C�� #�H�o� .�o��q���W�o�E� .� �}��N�N�s�$;�s��W�W�s�$;�<� ��H�H�U�O� � �x�x�%�,/�,3�-0�Y�Y�-?�-?�!A� B�� �(�(�#�555$�C�556 �C�y�R�Z�Z����b�i�i� @�@�H� �h�3�&J�c�&J�#J�#J� '�w� �8J�8J�K�
� � $��%�%�D�� � �S��[�)�2D�2D�$E�E�I� �+�+��]�);�Q�(?�?�K��c�2�:�:�o�&�+�+�B�H�H�5�C��c�3�w�<�=�(�)�C��&�&�&�&�1�1�#�2;�2=�8<� 2� >�C� �(�(�9�%�f�-�C� �q� ��#�t��t�)�n���C�y��T�]�]� 3�3� �3�i���J� �H�H�u�JM�)N�#�B�9�b�h�h�*?�#�)N� P� Q�M�557
�*�*�]��*�558?�C� �8�8�Y��V���/� 0�0��q
/�� %<��b *Os �L�)L�5L
559c � � U 4$ r r )r� r# r$ r� r� r� s r �&_unstructured_to_structured_dispatcherr . r� r! c �< � U R S:X a [ S5 eU R S nUS:X a [ S5 eUci Uc [ U5 Vs/ s H nSU 3PM560 nn[ R561 " U Vs/ s H owU R562 4PM snUS9n[
U5 n [ U 6 u p�n O�Ub [ S5 e[ R563 " U5 n[
U5 n [ U 5 S:X a / / / p�n564O [ U 6 u p�n U[ U 5 :w a [ S 5 eUnU( a UR ( d [ S5655 e[ [ U 5 5 Vs/ s H nSU 3PM566 nn[ R567 " UU568 V
s/ s H o�R569 U
R 4PM sn
S .5 n[ R " U 5 R U5 n [ R570 " UU571U UR S .5 nU R X�US
9n U R U5 S $ s snf s snf s snf s sn
f )ar 572Converts an n-D unstructured array into an (n-1)-D structured array.573 574The last dimension of the input array is converted into a structure, with575number of field-elements equal to the size of the last dimension of the576input array. By default all output fields have the input array's dtype, but577an output structured dtype with an equal number of fields-elements can be578supplied instead.579 580Nested fields, as well as each element of any subarray fields, all count581towards the number of field-elements.582 583Parameters584----------585arr : ndarray586 Unstructured array or dtype to convert.587dtype : dtype, optional588 The structured dtype of the output array589names : list of strings, optional590 If dtype is not supplied, this specifies the field names for the output591 dtype, in order. The field dtypes will be the same as the input array.592align : boolean, optional593 Whether to create an aligned memory layout.594copy : bool, optional595 See copy argument to `numpy.ndarray.astype`. If true, always return a596 copy. If false, and `dtype` requirements are satisfied, a view is597 returned.598casting : {'no', 'equiv', 'safe', 'same_kind', 'unsafe'}, optional599 See casting argument of `numpy.ndarray.astype`. Controls what kind of600 data casting may occur.601 602Returns603-------604structured : ndarray605 Structured array with fewer dimensions.606 607Examples608--------609>>> import numpy as np610 611>>> from numpy.lib import recfunctions as rfn612>>> dt = np.dtype([('a', 'i4'), ('b', 'f4,u2'), ('c', 'f4', 2)])613>>> a = np.arange(20).reshape((4,5))614>>> a615array([[ 0, 1, 2, 3, 4],616 [ 5, 6, 7, 8, 9],617 [10, 11, 12, 13, 14],618 [15, 16, 17, 18, 19]])619>>> rfn.unstructured_to_structured(a, dt)620array([( 0, ( 1., 2), [ 3., 4.]), ( 5, ( 6., 7), [ 8., 9.]),621 (10, (11., 12), [13., 14.]), (15, (16., 17), [18., 19.])],622 dtype=[('a', '<i4'), ('b', [('f0', '<f4'), ('f1', '<u2')]), ('c', '<f4', (2,))])623 624r z$arr must have at least one dimensionr� r z&last axis with size 0 is not supportedr3 r� z!don't supply both dtype and nameszVThe length of the last dimension of arr must be equal to the number of fields in dtypez'align was True but dtype is not alignedr� r� r� r� )r� r% r� r� rD r# r� r� r& r �isalignedstruct�ascontiguousarrayrj r� r� )r� r# r$ r� r� r� �n_elemr� r� r- r r� r� r� r r s r r r 2 s � �r �y�y�B���?�@�@�
�Y�Y�r�]�F�
��{�!�"J�K�K� �}� �=�&+�F�m�4�m��q���W�m�E�4��H�H�e� <�e��#�)�)�n�e� <�E�J� �(��3�� "�F�|� ��W� � � � @�A� A�������(��/�� �v�;�!� �#%�r�2��C��#&��<� �C�� �S��[� � � I� J�
J�� � ��2�2� � F�G� G� #�C��K� 0� 1� 0��q���W� 0�E� 1�625 �H�H�u�JM�)N�#�B�9�9�b�h�h�*?�#�)N� P� Q�M�626 �627 �628 �s�629#�630(�631(� �6327�C� �x�x�%�,/�,3�-6�-?�-?�!A� B�� �*�*�%�'�*�633B�C� �8�8�I� �v� &�&��Q 5�� <��,
2�� *Os �H634�4H� H�2!H635c � � U4$ r r )�funcr� s r � _apply_along_fields_dispatcherr � s � � �6�Mr! c �j � UR R c [ S5 e[ U5 nU " USS9$ )a 636Apply function 'func' as a reduction across fields of a structured array.637 638This is similar to `numpy.apply_along_axis`, but treats the fields of a639structured array as an extra axis. The fields are all first cast to a640common type following the type-promotion rules from `numpy.result_type`641applied to the field's dtypes.642 643Parameters644----------645func : function646 Function to apply on the "field" dimension. This function must647 support an `axis` argument, like `numpy.mean`, `numpy.sum`, etc.648arr : ndarray649 Structured array for which to apply func.650 651Returns652-------653out : ndarray654 Result of the reduction operation655 656Examples657--------658>>> import numpy as np659 660>>> from numpy.lib import recfunctions as rfn661>>> b = np.array([(1, 2, 5), (4, 5, 7), (7, 8 ,11), (10, 11, 12)],662... dtype=[('x', 'i4'), ('y', 'f4'), ('z', 'f8')])663>>> rfn.apply_along_fields(np.mean, b)664array([ 2.66666667, 5.33333333, 8.66666667, 11. ])665>>> rfn.apply_along_fields(np.mean, b[['x', 'z']])666array([ 3. , 5.5, 9. , 11. ])667 668r� r� )�axis)r# r$ r% r )r r� �uarrs r r r � s6 � �H �y�y��� ��9�:�:� %�c� *�D� ��2� � r! c � � X4$ r r )�dst�src�zero_unassigneds r �!_assign_fields_by_name_dispatcherr# � s � � �8�Or! c �� � U R R c XS' gU R R H< nX1R R ; a U( a SX' M) M+ [ X X U5 M> g)az 669Assigns values from one structured array to another by field name.670 671Normally in numpy >= 1.14, assignment of one structured array to another672copies fields "by position", meaning that the first field from the src is673copied to the first field of the dst, and so on, regardless of field name.674 675This function instead copies "by field name", such that fields in the dst676are assigned from the identically named field in the src. This applies677recursively for nested structures. This is how structure assignment worked678in numpy >= 1.6 to <= 1.13.679 680Parameters681----------682dst : ndarray683src : ndarray684 The source and destination arrays during assignment.685zero_unassigned : bool, optional686 If True, fields in the dst for which there was no matching687 field in the src are filled with the value 0 (zero). This688 was the behavior of numpy <= 1.13. If False, those fields689 are not modified.690N.r )r# r$ r )r r! r" r/ s r r r � s` � �4 �y�y��� ��C���� � ���� �y�y��� &� � �� � �
"�#�)�S�Y�"1�
3� r! c � � U 4$ r r )r� �required_dtypes r �_require_fields_dispatcherr' � s � � �8�Or! c �Z � [ R " U R US9n[ X 5 U$ )a� 691Casts a structured array to a new dtype using assignment by field-name.692 693This function assigns from the old to the new array by name, so the694value of a field in the output array is the value of the field with the695same name in the source array. This has the effect of creating a new696ndarray containing only the fields "required" by the required_dtype.697 698If a field name in the required_dtype does not exist in the699input array, that field is created and set to 0 in the output array.700 701Parameters702----------703a : ndarray704 array to cast705required_dtype : dtype706 datatype for output array707 708Returns709-------710out : ndarray711 array with the new dtype, with field values copied from the fields in712 the input array with the same name713 714Examples715--------716>>> import numpy as np717 718>>> from numpy.lib import recfunctions as rfn719>>> a = np.ones(4, dtype=[('a', 'i4'), ('b', 'f8'), ('c', 'u1')])720>>> rfn.require_fields(a, [('b', 'f4'), ('c', 'u1')])721array([(1., 1), (1., 1), (1., 1), (1., 1)],722 dtype=[('b', '<f4'), ('c', 'u1')])723>>> rfn.require_fields(a, [('b', 'f4'), ('newf', 'u1')])724array([(1., 0), (1., 0), (1., 0), (1., 0)],725 dtype=[('b', '<f4'), ('newf', 'u1')])726 727r� )rD r� r� r )r� r& �outs r r r � s& � �P
�(�(�5�;�;�n�7285�C��#�%� �Jr! c � � U $ r r )�arraysrv ro rp � autoconverts r �_stack_arrays_dispatcherr- & s � � �Mr! c �< � [ U [ R 5 ( a U $ [ U 5 S:X a U S $ U Vs/ s H' n[ R " U5 R 5 PM) nnU Vs/ s H n[ U5 PM nnU Vs/ s H oUR PM nnU V s/ s H o�R PM n729n US n [ U 5 n U V
V s/ s H u p�U
PM nn
n USS H� n[ U5 H} u nnUU; a& U R UU45 UR U5 M2 UR U5 nU U u nnU( a U[ UU5 4U U' Me UU:w d Mm [ SU SU S35 e M� [ U 5 S:X a [ R " U5 nO�[ R " [ R " U5 4U 5 n[ R" " [ R$ SU4 5 n/ n[' XjUSS USS 5 Hc u p]nnUR R nUc UUS[ U5 3 UU& M6 U
H' nUU UU UU& UU; d M UR U5 M) Me [) [+ UU5 X#S 9$ s snf s snf s snf s sn f s sn n
f )730a� 731Superposes arrays fields by fields732 733Parameters734----------735arrays : array or sequence736 Sequence of input arrays.737defaults : dictionary, optional738 Dictionary mapping field names to the corresponding default values.739usemask : {True, False}, optional740 Whether to return a MaskedArray (or MaskedRecords is741 `asrecarray==True`) or a ndarray.742asrecarray : {False, True}, optional743 Whether to return a recarray (or MaskedRecords if `usemask==True`)744 or just a flexible-type ndarray.745autoconvert : {False, True}, optional746 Whether automatically cast the type of the field to the maximum.747 748Examples749--------750>>> import numpy as np751>>> from numpy.lib import recfunctions as rfn752>>> x = np.array([1, 2,])753>>> rfn.stack_arrays(x) is x754True755>>> z = np.array([('A', 1), ('B', 2)], dtype=[('A', '|S3'), ('B', float)])756>>> zz = np.array([('a', 10., 100.), ('b', 20., 200.), ('c', 30., 300.)],757... dtype=[('A', '|S3'), ('B', np.double), ('C', np.double)])758>>> test = rfn.stack_arrays((z,zz))759>>> test760masked_array(data=[(b'A', 1.0, --), (b'B', 2.0, --), (b'a', 10.0, 100.0),761 (b'b', 20.0, 200.0), (b'c', 30.0, 300.0)],762 mask=[(False, False, True), (False, False, True),763 (False, False, False), (False, False, False),764 (False, False, False)],765 fill_value=(b'N/A', 1e+20, 1e+20),766 dtype=[('A', 'S3'), ('B', '<f8'), ('C', '<f8')])767 768rC r NzIncompatible type 'z' <> '�'r� r3 r� )rT rD r� r& r� r� r# r$ r4 r6 �indexr� � TypeErrorrh � concatenater� r �cumsum�r_r� rq ry )r+ rv ro rp r, rG rE �nrecordsr> r� �fldnames�dtype_l�newdescrr� r$ �dtype_n�fname�fdtype�nameidxrA �cdtyper r� �seenr� �jr/ s r r r + sq � �T �&�"�*�*�%�%��
� �V��� ��a�y��39�:�6�a�� � �q�!�'�'�)�6�I�:� )�*� �1��A�� �H�*� (�
)�y�!�g�g�y�F�
)�!'�(��A����H�(��Q�i�G� �g�&�H�#� $�8�4�1�Q�8�E� $��!�"�:�� +�G� 4�M�E�6��E�!�������0�� � �U�#� �+�+�e�,�� $�W� -� ��6� �).��F�F�0C�(D�H�W�%��v�%�#�&9�&����x�q�$Q�R�R� 5� � �8�}���� � � �*�� � � ����x� 0� 2�H�=�����2�5�5��H��-�.���� � �V�C�R�[�&���*� M�L�Q�1�a��G�G�M�M�E��}�/0���3�t�9�+��'��!�,� �D�()�$��F�4�L��1�%��4�'� � � �D�)� � N� �}�V�X�6� &� ?� ?��M ;��*��
)��(��
%s �.J�.J � J�$J�Jc � � U 4$ r r )rG �key�769ignoremask� return_indexs r �_find_duplicates_dispatcherrD � s � �
�4�Kr! c � � [ R " U 5 R 5 n [ U R 5 nU nU( a XA H nXV nM XQ nUR 5 nXW nUR
5 n U SS U SS :H n770U( a UR n SX�SS ' [ R " S/U77145 n772U773SS U774SS - U775SS& X U776 n U( a X�U777 4$ U $ )a� 778Find the duplicates in a structured array along a given key779 780Parameters781----------782a : array-like783 Input array784key : {string, None}, optional785 Name of the fields along which to check the duplicates.786 If None, the search is performed by records787ignoremask : {True, False}, optional788 Whether masked data should be discarded or considered as duplicates.789return_index : {False, True}, optional790 Whether to return the indices of the duplicated values.791 792Examples793--------794>>> import numpy as np795>>> from numpy.lib import recfunctions as rfn796>>> ndtype = [('a', int)]797>>> a = np.ma.array([1, 1, 1, 2, 2, 3, 3],798... mask=[0, 0, 1, 0, 0, 0, 1]).view(ndtype)799>>> rfn.find_duplicates(a, ignoremask=True, return_index=True)800(masked_array(data=[(1,), (1,), (2,), (2,)],801 mask=[(False,), (False,), (False,), (False,)],802 fill_value=(999999,),803 dtype=[('a', '<i8')]), array([0, 1, 3, 4]))804Nr� rC F) rD r� r� r r# �argsortrm �805recordmaskr2 )
rG rA rB rC r- r� r3 �sortidx�806sortedbase�807sorteddata�flag�808sortedmask�809duplicatess
r r r � s� � �< �
�
�a�� � � �A�
����
(�F� �D�810���A��7�D� ��y���l�l�n�G�� �J��"�"�$�J� �s��O�z�!�"�~� -�D�� �*�*�811� $����^� �
�>�>�E�7�D�/� *�D��S�b� �D���H�$�D��"�I���D�!�J���D�M�*�*��r! c � � X4$ r r ) rA �r1�r2�jointype� r1postfix� r2postfixrv ro rp s r �_join_by_dispatcherrT � � � �
�8�Or! c �812 ^ � US; a [ SU- 5 e[ T [ 5 ( a T 4m [ [ T 5 5 [ T 5 :w a- [ U 4S j[
T 5 5 5 n [ SU < 35 eT HS n813X�R R ; a [ SU814< 35 eX�R R ; d MF [ SU815< 35 e UR 5 nUR 5 n[ U5 [ U5 p�UR R UR R p�[ U
5 [ U5 - [ T 5 -816 nU( a U( d U( d SnUS-
n[ U5 eU
Vs/ s H
nUT ; d M UPM nn[ UU5 n[ UU5 n[ R " UU45 nUR T S 9nUU n[ R " S817/US S US S
:H 45 nUS S US S
- US S
& UU nUUU : nUUU :� U -818 n[ U5 [ U5 n nUS:X a Su n n O�US:X ab UU) n [ R " UU U U : 45 n[ R " UU U U :� U -819 45 n[ U5 U-820 [ U5 U -821 n n O:US:X a4 UU) n [ R " UU U U : 45 n[ U5 U-822 Sn n UU UU n!n [ UR 5 n"[ UR 5 H! u n#n$U#T ; d M U"R! U#U$45 M# [ UR 5 Hd u n#n$U" V823V%s/ s H u n824n%U825PM826 n&n827n% U&R# U#5 n'U"U' u n(n)U#T ; a U#[% U$U)5 4U"U'' MP U#U- U)4U#U- U$4/U"U'U'S - & Mf [ R " U"5 n"[% UU 5 n*[ R& " U*W - W - 4U"S9n+U+R R n&U
HF n,U U, n-U,U&; d U,U; a U( d U,T ; a U,U-
n,U+U, n.U-S U U.S U& US; d M; U-US U.U*U*U - & MH U HM n,U!U, n-U,U&; d U,U
; a U( d U,T ; a U,U-
n,U+U, n.U-S U U.S U & US:X d M; U ( d MD U-U S U.U * S & MO U+R) T S 9 XxS.n/[+ [- U+U5 40 U/D6$ s snf s sn%n828f ! [ a U"R! U#U$45 GM� f = f)a� 829Join arrays `r1` and `r2` on key `key`.830 831The key should be either a string or a sequence of string corresponding832to the fields used to join the array. An exception is raised if the833`key` field cannot be found in the two input arrays. Neither `r1` nor834`r2` should have any duplicates along `key`: the presence of duplicates835will make the output quite unreliable. Note that duplicates are not836looked for by the algorithm.837 838Parameters839----------840key : {string, sequence}841 A string or a sequence of strings corresponding to the fields used842 for comparison.843r1, r2 : arrays844 Structured arrays.845jointype : {'inner', 'outer', 'leftouter'}, optional846 If 'inner', returns the elements common to both r1 and r2.847 If 'outer', returns the common elements as well as the elements of848 r1 not in r2 and the elements of not in r2.849 If 'leftouter', returns the common elements and the elements of r1850 not in r2.851r1postfix : string, optional852 String appended to the names of the fields of r1 that are present853 in r2 but absent of the key.854r2postfix : string, optional855 String appended to the names of the fields of r2 that are present856 in r1 but absent of the key.857defaults : {dictionary}, optional858 Dictionary mapping field names to the corresponding default values.859usemask : {True, False}, optional860 Whether to return a MaskedArray (or MaskedRecords is861 `asrecarray==True`) or a ndarray.862asrecarray : {False, True}, optional863 Whether to return a recarray (or MaskedRecords if `usemask==True`)864 or just a flexible-type ndarray.865 866Notes867-----868* The output is sorted along the key.869* A temporary array is formed by dropping the fields not in the key for870 the two arrays and concatenating the result. This array is then871 sorted, and the common entries selected. The output is constructed by872 filling the fields with the selected entries. Matching is not873 preserved if there are some duplicates...874 875)�inner�outer� leftouterzWThe 'jointype' argument should be in 'inner', 'outer' or 'leftouter' (got '%s' instead)c 3 �H ># � U H u pUTUS - S ; d M Uv � M g7f)rC Nr )r. r� �xrA s �r r0 �join_by.<locals>.<genexpr> s& �� � �D� ���1��A��E�F� �3C�1�1� �s �"� "zduplicate join key zr1 does not have key field zr2 does not have key field z8r1 and r2 contain common names, r1postfix and r2postfix zcan't both be empty)�orderFrC Nr� rW )r r rX rY r r� )rX rY r� )r% rT r] r& r� �next� enumerater# r$ r� r� rh r2 rF rD r4 r6 r0 r� r� �sortrq ry )0rA rO rP rQ rR rS rv ro rp �dupr/ �nb1�nb2�r1names�r2names�876collisionsr� r� �key1�r1k�r2k�aux�idx_sort�flag_in�idx_in�idx_1�idx_2�r1cmn�r2cmn�r1spc�r2spc�idx_out�s1�s2r> r: r; r# r$ r< rA r= �cmnr r3 �selectedr) �kwargss0 ` r r r � s[ �� �h �6�6��<�>F�G�� �877 �#�s����f�� �3�s�8�}��C�� ��D��3� �D�D���.�s�g�6�7�7��� �x�x�~�~� %� � :�4�(� C�D� D� �x�x�~�~� %� � :�4�(� C�D� D� � 878 ����B� ����B��b�'�3�r�7�#� �(�(�.�.�"�(�(�.�.�g� �g�,��W� �-��S��9�J��9� �H�� �$�$����o� � � +�w�!�!�s�(�A�w�D� +�879�r�4�880 �C�881�r�4�882 �C�
�.�.�#�s��883$�C��{�{��{�%�H�884
�h�-�C� �n�n�u�g�s�1�2�w�#�c�r�(�':� ;�<�G��1�2�;���"� �-�G�C�R�L�
�g�
�F� �F�S�L� "�E� �F�c�M� #�c� )�E��%�j�#�e�*�E�U��7�� ���� �W� ��G�8�$������w��#�
�'?� @�A������w��3��'@�3�'F� G�H�� �e�*�u�,�c�%�j�5�.@���� �[� ��G�8�$������w��#�
�'?� @�A�� �e�*�u�,�a����5� �2�e�9��R� �C�I�I�
&�F� (����1�
��v� �� � �M�M�5�&�/� *� 2�885 (����1�
��v� *0�0��+�$�����0� ��k�k�%�(�G� �w��I�A�v���|�#(�#�f�f�*=�">��w��886 �Y�&��/��Y�&��/�/��w�w��{�+�% 2�. �X�X�f�
�F� �e�U�887�C�
�]�]�C�%�K�%� /�1��
@�F� �L�L� � �E�
���a�5�� �E�>�a�7�l�9��#� �
��N�A���)��"�6�E�*����� �-� -�'/���'7�G�C��e� � $� � ���a�5�� �E�>�a�7�l�9��#� �
��N�A���)��"�6�E�*����� �� �U�U� '��� /�G�U�F�G� � � �K�K�c�K�� �
;�F� �}�V�X�6� A�&� A�A��s ,��V 1�� � +� �M�M�5�&�/� *� *� +�s$ �888S�S�"S �5S&�& T�Tc � � X4$ r r )rA rO rP rQ rR rS rv s r �_rec_join_dispatcherr{ { rU r! c �, � X4UUSSS.n[ XU40 UD6$ )z�889Join arrays `r1` and `r2` on keys.890Alternative to join_by, that always returns a np.recarray.891 892See Also893--------894join_by : equivalent function895FT)rQ rR rS rv ro rp )r )rA rO rP rQ rR rS rv ry s r r r � s* � � #�� &�5��N�F� �3�B� )�&� )�)r! )F)NN)NT)TFr )NNNN)r� FFF)Nr� TF)FF)r )NNN)NF�unsafe)NNNNN)NNFFr} )T)NTFF)NTF)NNNNNN)rW �1�2NTF)rW r~ r N)?�__doc__ra �numpyrD �numpy.marh �numpy.ma.mrecords�mrecordsrk �numpy._core.overridesr �numpy.lib._iotoolsr �__all__r r r4 r r
r896 rH rK r rV r^ rf rq ry r{ r r� r r� r� r r� r r� r r� r r� r r� r� r� r r r r r r# r r' r r- r rD r rT r r{ r r r! r �<module>r� s ��� � � � � � 9� .� ��� �:�;�$� <�$�N 897�@ �B �B �6
� 8� -�`898�
� "�0�" � BF�6:��899 �1�2�38�+0�E� 3�E�P� �0�1�:G� 2�:G�zG�2� �4�5�I� 6�I�� �2�3�% � 4�% �P 9=�HL�� �2�3�,0�:?�CG� 4�CG�L�900 �6�7� 9� 8� 9�D� �2�3�Q"� 4�Q"�f#�J*�Z BF�37�� �?�@�1� A�1�D CG�JN�� �?�@�BG�3;�i'� A�i'�V� �7�8�' � 9�' �V� �:�;�#3� <�#3�J� �3�4�)� 5�)�X =A�:>��901 �1�2�AF� "�T?� 3�T?�p 48��902 �4�5�7� 6�7�v ?C�04�� �,�-�DG�49�nB� .�nB�d ?C��� �-�.�EH�� *� /� *� r!