1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
|
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<meta name="generator" content="Docutils 0.4: http://docutils.sourceforge.net/" />
<title>numpy.i: a SWIG Interface File for NumPy</title>
<meta name="author" content="Bill Spotz" />
<meta name="date" content="13 April, 2007" />
<style type="text/css">
/*
:Author: David Goodger
:Contact: goodger@users.sourceforge.net
:Date: $Date: 2005-12-18 01:56:14 +0100 (Sun, 18 Dec 2005) $
:Revision: $Revision: 4224 $
:Copyright: This stylesheet has been placed in the public domain.
Default cascading style sheet for the HTML output of Docutils.
See http://docutils.sf.net/docs/howto/html-stylesheets.html for how to
customize this style sheet.
*/
/* used to remove borders from tables and images */
.borderless, table.borderless td, table.borderless th {
border: 0 }
table.borderless td, table.borderless th {
/* Override padding for "table.docutils td" with "! important".
The right padding separates the table cells. */
padding: 0 0.5em 0 0 ! important }
.first {
/* Override more specific margin styles with "! important". */
margin-top: 0 ! important }
.last, .with-subtitle {
margin-bottom: 0 ! important }
.hidden {
display: none }
a.toc-backref {
text-decoration: none ;
color: black }
blockquote.epigraph {
margin: 2em 5em ; }
dl.docutils dd {
margin-bottom: 0.5em }
/* Uncomment (and remove this text!) to get bold-faced definition list terms
dl.docutils dt {
font-weight: bold }
*/
div.abstract {
margin: 2em 5em }
div.abstract p.topic-title {
font-weight: bold ;
text-align: center }
div.admonition, div.attention, div.caution, div.danger, div.error,
div.hint, div.important, div.note, div.tip, div.warning {
margin: 2em ;
border: medium outset ;
padding: 1em }
div.admonition p.admonition-title, div.hint p.admonition-title,
div.important p.admonition-title, div.note p.admonition-title,
div.tip p.admonition-title {
font-weight: bold ;
font-family: sans-serif }
div.attention p.admonition-title, div.caution p.admonition-title,
div.danger p.admonition-title, div.error p.admonition-title,
div.warning p.admonition-title {
color: red ;
font-weight: bold ;
font-family: sans-serif }
/* Uncomment (and remove this text!) to get reduced vertical space in
compound paragraphs.
div.compound .compound-first, div.compound .compound-middle {
margin-bottom: 0.5em }
div.compound .compound-last, div.compound .compound-middle {
margin-top: 0.5em }
*/
div.dedication {
margin: 2em 5em ;
text-align: center ;
font-style: italic }
div.dedication p.topic-title {
font-weight: bold ;
font-style: normal }
div.figure {
margin-left: 2em ;
margin-right: 2em }
div.footer, div.header {
clear: both;
font-size: smaller }
div.line-block {
display: block ;
margin-top: 1em ;
margin-bottom: 1em }
div.line-block div.line-block {
margin-top: 0 ;
margin-bottom: 0 ;
margin-left: 1.5em }
div.sidebar {
margin-left: 1em ;
border: medium outset ;
padding: 1em ;
background-color: #ffffee ;
width: 40% ;
float: right ;
clear: right }
div.sidebar p.rubric {
font-family: sans-serif ;
font-size: medium }
div.system-messages {
margin: 5em }
div.system-messages h1 {
color: red }
div.system-message {
border: medium outset ;
padding: 1em }
div.system-message p.system-message-title {
color: red ;
font-weight: bold }
div.topic {
margin: 2em }
h1.section-subtitle, h2.section-subtitle, h3.section-subtitle,
h4.section-subtitle, h5.section-subtitle, h6.section-subtitle {
margin-top: 0.4em }
h1.title {
text-align: center }
h2.subtitle {
text-align: center }
hr.docutils {
width: 75% }
img.align-left {
clear: left }
img.align-right {
clear: right }
ol.simple, ul.simple {
margin-bottom: 1em }
ol.arabic {
list-style: decimal }
ol.loweralpha {
list-style: lower-alpha }
ol.upperalpha {
list-style: upper-alpha }
ol.lowerroman {
list-style: lower-roman }
ol.upperroman {
list-style: upper-roman }
p.attribution {
text-align: right ;
margin-left: 50% }
p.caption {
font-style: italic }
p.credits {
font-style: italic ;
font-size: smaller }
p.label {
white-space: nowrap }
p.rubric {
font-weight: bold ;
font-size: larger ;
color: maroon ;
text-align: center }
p.sidebar-title {
font-family: sans-serif ;
font-weight: bold ;
font-size: larger }
p.sidebar-subtitle {
font-family: sans-serif ;
font-weight: bold }
p.topic-title {
font-weight: bold }
pre.address {
margin-bottom: 0 ;
margin-top: 0 ;
font-family: serif ;
font-size: 100% }
pre.literal-block, pre.doctest-block {
margin-left: 2em ;
margin-right: 2em ;
background-color: #eeeeee }
span.classifier {
font-family: sans-serif ;
font-style: oblique }
span.classifier-delimiter {
font-family: sans-serif ;
font-weight: bold }
span.interpreted {
font-family: sans-serif }
span.option {
white-space: nowrap }
span.pre {
white-space: pre }
span.problematic {
color: red }
span.section-subtitle {
/* font-size relative to parent (h1..h6 element) */
font-size: 80% }
table.citation {
border-left: solid 1px gray;
margin-left: 1px }
table.docinfo {
margin: 2em 4em }
table.docutils {
margin-top: 0.5em ;
margin-bottom: 0.5em }
table.footnote {
border-left: solid 1px black;
margin-left: 1px }
table.docutils td, table.docutils th,
table.docinfo td, table.docinfo th {
padding-left: 0.5em ;
padding-right: 0.5em ;
vertical-align: top }
table.docutils th.field-name, table.docinfo th.docinfo-name {
font-weight: bold ;
text-align: left ;
white-space: nowrap ;
padding-left: 0 }
h1 tt.docutils, h2 tt.docutils, h3 tt.docutils,
h4 tt.docutils, h5 tt.docutils, h6 tt.docutils {
font-size: 100% }
tt.docutils {
background-color: #eeeeee }
ul.auto-toc {
list-style-type: none }
</style>
</head>
<body>
<div class="document" id="numpy-i-a-swig-interface-file-for-numpy">
<h1 class="title">numpy.i: a SWIG Interface File for NumPy</h1>
<table class="docinfo" frame="void" rules="none">
<col class="docinfo-name" />
<col class="docinfo-content" />
<tbody valign="top">
<tr><th class="docinfo-name">Author:</th>
<td>Bill Spotz</td></tr>
<tr class="field"><th class="docinfo-name">Institution:</th><td class="field-body">Sandia National Laboratories</td>
</tr>
<tr><th class="docinfo-name">Date:</th>
<td>13 April, 2007</td></tr>
</tbody>
</table>
<div class="contents topic">
<p class="topic-title first"><a id="contents" name="contents">Contents</a></p>
<ul class="simple">
<li><a class="reference" href="#introduction" id="id1" name="id1">Introduction</a></li>
<li><a class="reference" href="#using-numpy-i" id="id2" name="id2">Using numpy.i</a></li>
<li><a class="reference" href="#available-typemaps" id="id3" name="id3">Available Typemaps</a><ul>
<li><a class="reference" href="#input-arrays" id="id4" name="id4">Input Arrays</a></li>
<li><a class="reference" href="#in-place-arrays" id="id5" name="id5">In-Place Arrays</a></li>
<li><a class="reference" href="#argout-arrays" id="id6" name="id6">Argout Arrays</a></li>
<li><a class="reference" href="#output-arrays" id="id7" name="id7">Output Arrays</a></li>
<li><a class="reference" href="#other-common-types-bool" id="id8" name="id8">Other Common Types: bool</a></li>
<li><a class="reference" href="#other-common-types-complex" id="id9" name="id9">Other Common Types: complex</a></li>
</ul>
</li>
<li><a class="reference" href="#helper-functions" id="id10" name="id10">Helper Functions</a><ul>
<li><a class="reference" href="#macros" id="id11" name="id11">Macros</a></li>
<li><a class="reference" href="#routines" id="id12" name="id12">Routines</a></li>
</ul>
</li>
<li><a class="reference" href="#beyond-the-provided-typemaps" id="id13" name="id13">Beyond the Provided Typemaps</a><ul>
<li><a class="reference" href="#a-common-example" id="id14" name="id14">A Common Example</a></li>
<li><a class="reference" href="#other-situations" id="id15" name="id15">Other Situations</a></li>
<li><a class="reference" href="#a-final-note" id="id16" name="id16">A Final Note</a></li>
</ul>
</li>
<li><a class="reference" href="#summary" id="id17" name="id17">Summary</a></li>
<li><a class="reference" href="#acknowledgements" id="id18" name="id18">Acknowledgements</a></li>
</ul>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id1" id="introduction" name="introduction">Introduction</a></h1>
<p>The Simple Wrapper and Interface Generator (or <a class="reference" href="http://www.swig.org">SWIG</a>) is a powerful tool for generating wrapper
code for interfacing to a wide variety of scripting languages.
<a class="reference" href="http://www.swig.org">SWIG</a> can parse header files, and using only the code prototypes,
create an interface to the target language. But <a class="reference" href="http://www.swig.org">SWIG</a> is not
omnipotent. For example, it cannot know from the prototype:</p>
<pre class="literal-block">
double rms(double* seq, int n);
</pre>
<p>what exactly <tt class="docutils literal"><span class="pre">seq</span></tt> is. Is it a single value to be altered in-place?
Is it an array, and if so what is its length? Is it input-only?
Output-only? Input-output? <a class="reference" href="http://www.swig.org">SWIG</a> cannot determine these details,
and does not attempt to do so.</p>
<p>Making an educated guess, humans can conclude that this is probably a
routine that takes an input-only array of length <tt class="docutils literal"><span class="pre">n</span></tt> of <tt class="docutils literal"><span class="pre">double</span></tt>
values called <tt class="docutils literal"><span class="pre">seq</span></tt> and returns the root mean square. The default
behavior of <a class="reference" href="http://www.swig.org">SWIG</a>, however, will be to create a wrapper function
that compiles, but is nearly impossible to use from the scripting
language in the way the C routine was intended.</p>
<p>For <a class="reference" href="http://www.python.org">python</a>, the preferred way of handling
contiguous (or technically, <em>strided</em>) blocks of homogeneous data is
with the module <a class="reference" href="http://numpy.scipy.org">NumPy</a>, which provides full
object-oriented access to arrays of data. Therefore, the most logical
<a class="reference" href="http://www.python.org">python</a> interface for the <tt class="docutils literal"><span class="pre">rms</span></tt> function would be (including doc
string):</p>
<pre class="literal-block">
def rms(seq):
"""
rms: return the root mean square of a sequence
rms(numpy.ndarray) -> double
rms(list) -> double
rms(tuple) -> double
"""
</pre>
<p>where <tt class="docutils literal"><span class="pre">seq</span></tt> would be a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array of <tt class="docutils literal"><span class="pre">double</span></tt> values, and its
length <tt class="docutils literal"><span class="pre">n</span></tt> would be extracted from <tt class="docutils literal"><span class="pre">seq</span></tt> internally before being
passed to the C routine. Even better, since <a class="reference" href="http://numpy.scipy.org">NumPy</a> supports
construction of arrays from arbitrary <a class="reference" href="http://www.python.org">python</a> sequences, <tt class="docutils literal"><span class="pre">seq</span></tt>
itself could be a nearly arbitrary sequence (so long as each element
can be converted to a <tt class="docutils literal"><span class="pre">double</span></tt>) and the wrapper code would
internally convert it to a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array before extracting its data
and length.</p>
<p><a class="reference" href="http://www.swig.org">SWIG</a> allows these types of conversions to be defined via a
mechanism called typemaps. This document provides information on how
to use <tt class="docutils literal"><span class="pre">numpy.i</span></tt>, a <a class="reference" href="http://www.swig.org">SWIG</a> interface file that defines a series of
typemaps intended to make the type of array-related conversions
described above relatively simple to implement. For example, suppose
that the <tt class="docutils literal"><span class="pre">rms</span></tt> function prototype defined above was in a header file
named <tt class="docutils literal"><span class="pre">rms.h</span></tt>. To obtain the <a class="reference" href="http://www.python.org">python</a> interface discussed above,
your <a class="reference" href="http://www.swig.org">SWIG</a> interface file would need the following:</p>
<pre class="literal-block">
%{
#define SWIG_FILE_WITH_INIT
#include "rms.h"
%}
%include "numpy.i"
%init %{
import_array();
%}
%apply (double* IN_ARRAY1, int DIM1) {(double* seq, int n)};
%include "rms.h"
</pre>
<p>Typemaps are keyed off a list of one or more function arguments,
either by type or by type and name. We will refer to such lists as
<em>signatures</em>. One of the many typemaps defined by <tt class="docutils literal"><span class="pre">numpy.i</span></tt> is used
above and has the signature <tt class="docutils literal"><span class="pre">(double*</span> <span class="pre">IN_ARRAY1,</span> <span class="pre">int</span> <span class="pre">DIM1)</span></tt>. The
argument names are intended to suggest that the <tt class="docutils literal"><span class="pre">double*</span></tt> argument
is an input array of one dimension and that the <tt class="docutils literal"><span class="pre">int</span></tt> represents
that dimension. This is precisely the pattern in the <tt class="docutils literal"><span class="pre">rms</span></tt>
prototype.</p>
<p>Most likely, no actual prototypes to be wrapped will have the argument
names <tt class="docutils literal"><span class="pre">IN_ARRAY1</span></tt> and <tt class="docutils literal"><span class="pre">DIM1</span></tt>. We use the <tt class="docutils literal"><span class="pre">%apply</span></tt> directive to
apply the typemap for one-dimensional input arrays of type <tt class="docutils literal"><span class="pre">double</span></tt>
to the actual prototype used by <tt class="docutils literal"><span class="pre">rms</span></tt>. Using <tt class="docutils literal"><span class="pre">numpy.i</span></tt>
effectively, therefore, requires knowing what typemaps are available
and what they do.</p>
<p>A <a class="reference" href="http://www.swig.org">SWIG</a> interface file that includes the <a class="reference" href="http://www.swig.org">SWIG</a> directives given
above will produce wrapper code that looks something like:</p>
<pre class="literal-block">
1 PyObject *_wrap_rms(PyObject *args) {
2 PyObject *resultobj = 0;
3 double *arg1 = (double *) 0 ;
4 int arg2 ;
5 double result;
6 PyArrayObject *array1 = NULL ;
7 int is_new_object1 = 0 ;
8 PyObject * obj0 = 0 ;
9
10 if (!PyArg_ParseTuple(args,(char *)"O:rms",&obj0)) SWIG_fail;
11 {
12 array1 = obj_to_array_contiguous_allow_conversion(
13 obj0, NPY_DOUBLE, &is_new_object1);
14 npy_intp size[1] = {
15 -1
16 };
17 if (!array1 || !require_dimensions(array1, 1) ||
18 !require_size(array1, size, 1)) SWIG_fail;
19 arg1 = (double*) array1->data;
20 arg2 = (int) array1->dimensions[0];
21 }
22 result = (double)rms(arg1,arg2);
23 resultobj = SWIG_From_double((double)(result));
24 {
25 if (is_new_object1 && array1) Py_DECREF(array1);
26 }
27 return resultobj;
28 fail:
29 {
30 if (is_new_object1 && array1) Py_DECREF(array1);
31 }
32 return NULL;
33 }
</pre>
<p>The typemaps from <tt class="docutils literal"><span class="pre">numpy.i</span></tt> are responsible for the following lines
of code: 12--20, 25 and 30. Line 10 parses the input to the <tt class="docutils literal"><span class="pre">rms</span></tt>
function. From the format string <tt class="docutils literal"><span class="pre">"O:rms"</span></tt>, we can see that the
argument list is expected to be a single <a class="reference" href="http://www.python.org">python</a> object (specified
by the <tt class="docutils literal"><span class="pre">O</span></tt> before the colon) and whose pointer is stored in
<tt class="docutils literal"><span class="pre">obj0</span></tt>. A number of functions, supplied by <tt class="docutils literal"><span class="pre">numpy.i</span></tt>, are called
to make and check the (possible) conversion from a generic <a class="reference" href="http://www.python.org">python</a>
object to a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array. These functions are explained in the
section <a class="reference" href="#helper-functions">Helper Functions</a>, but hopefully their names are
self-explanatory. At line 12 we use <tt class="docutils literal"><span class="pre">obj0</span></tt> to construct a <a class="reference" href="http://numpy.scipy.org">NumPy</a>
array. At line 17, we check the validity of the result: that it is
non-null and that it has a single dimension of arbitrary length. Once
these states are verified, we extract the data buffer and length in
lines 19 and 20 so that we can call the underlying C function at line
22. Line 25 performs memory management for the case where we have
created a new array that is no longer needed.</p>
<p>This code has a significant amount of error handling. Note the
<tt class="docutils literal"><span class="pre">SWIG_fail</span></tt> is a macro for <tt class="docutils literal"><span class="pre">goto</span> <span class="pre">fail</span></tt>, refering to the label at
line 28. If the user provides the wrong number of arguments, this
will be caught at line 10. If construction of the <a class="reference" href="http://numpy.scipy.org">NumPy</a> array
fails or produces an array with the wrong number of dimensions, these
errors are caught at line 17. And finally, if an error is detected,
memory is still managed correctly at line 30.</p>
<p>Note that if the C function signature was in a different order:</p>
<pre class="literal-block">
double rms(int n, double* seq);
</pre>
<p>that <a class="reference" href="http://www.swig.org">SWIG</a> would not match the typemap signature given above with
the argument list for <tt class="docutils literal"><span class="pre">rms</span></tt>. Fortunately, <tt class="docutils literal"><span class="pre">numpy.i</span></tt> has a set of
typemaps with the data pointer given last:</p>
<pre class="literal-block">
%apply (int DIM1, double* IN_ARRAY1) {(int n, double* seq)};
</pre>
<p>This simply has the effect of switching the definitions of <tt class="docutils literal"><span class="pre">arg1</span></tt>
and <tt class="docutils literal"><span class="pre">arg2</span></tt> in lines 3 and 4 of the generated code above, and their
assignments in lines 19 and 20.</p>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id2" id="using-numpy-i" name="using-numpy-i">Using numpy.i</a></h1>
<p>The <tt class="docutils literal"><span class="pre">numpy.i</span></tt> file is currently located in the <tt class="docutils literal"><span class="pre">numpy/docs/swig</span></tt>
sub-directory under the <tt class="docutils literal"><span class="pre">numpy</span></tt> installation directory. Typically,
you will want to copy it to the directory where you are developing
your wrappers. If it is ever adopted by <a class="reference" href="http://www.swig.org">SWIG</a> developers, then it
will be installed in a standard place where <a class="reference" href="http://www.swig.org">SWIG</a> can find it.</p>
<p>A simple module that only uses a single <a class="reference" href="http://www.swig.org">SWIG</a> interface file should
include the following:</p>
<pre class="literal-block">
%{
#define SWIG_FILE_WITH_INIT
%}
%include "numpy.i"
%init %{
import_array();
%}
</pre>
<p>Within a compiled <a class="reference" href="http://www.python.org">python</a> module, <tt class="docutils literal"><span class="pre">import_array()</span></tt> should only get
called once. This could be in a C/C++ file that you have written and
is linked to the module. If this is the case, then none of your
interface files should <tt class="docutils literal"><span class="pre">#define</span> <span class="pre">SWIG_FILE_WITH_INIT</span></tt> or call
<tt class="docutils literal"><span class="pre">import_array()</span></tt>. Or, this initialization call could be in a
wrapper file generated by <a class="reference" href="http://www.swig.org">SWIG</a> from an interface file that has the
<tt class="docutils literal"><span class="pre">%init</span></tt> block as above. If this is the case, and you have more than
one <a class="reference" href="http://www.swig.org">SWIG</a> interface file, then only one interface file should
<tt class="docutils literal"><span class="pre">#define</span> <span class="pre">SWIG_FILE_WITH_INIT</span></tt> and call <tt class="docutils literal"><span class="pre">import_array()</span></tt>.</p>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id3" id="available-typemaps" name="available-typemaps">Available Typemaps</a></h1>
<p>The typemap directives provided by <tt class="docutils literal"><span class="pre">numpy.i</span></tt> for arrays of different
data types, say <tt class="docutils literal"><span class="pre">double</span></tt> and <tt class="docutils literal"><span class="pre">int</span></tt>, and dimensions of different
types, say <tt class="docutils literal"><span class="pre">int</span></tt> or <tt class="docutils literal"><span class="pre">long</span></tt>, are identical to one another except
for the C and <a class="reference" href="http://numpy.scipy.org">NumPy</a> type specifications. The typemaps are
therefore implemented (typically behind the scenes) via a macro:</p>
<pre class="literal-block">
%numpy_typemaps(DATA_TYPE, DATA_TYPECODE, DIM_TYPE)
</pre>
<p>that can be invoked for appropriate <tt class="docutils literal"><span class="pre">(DATA_TYPE,</span> <span class="pre">DATA_TYPECODE,</span>
<span class="pre">DIM_TYPE)</span></tt> triplets. For example:</p>
<pre class="literal-block">
%numpy_typemaps(double, NPY_DOUBLE, int)
%numpy_typemaps(int, NPY_INT , int)
</pre>
<p>The <tt class="docutils literal"><span class="pre">numpy.i</span></tt> interface file uses the <tt class="docutils literal"><span class="pre">%numpy_typemaps</span></tt> macro to
implement typemaps for the following C data types and <tt class="docutils literal"><span class="pre">int</span></tt>
dimension types:</p>
<blockquote>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">signed</span> <span class="pre">char</span></tt></li>
<li><tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">char</span></tt></li>
<li><tt class="docutils literal"><span class="pre">short</span></tt></li>
<li><tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">short</span></tt></li>
<li><tt class="docutils literal"><span class="pre">int</span></tt></li>
<li><tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">int</span></tt></li>
<li><tt class="docutils literal"><span class="pre">long</span></tt></li>
<li><tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">long</span></tt></li>
<li><tt class="docutils literal"><span class="pre">long</span> <span class="pre">long</span></tt></li>
<li><tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">long</span> <span class="pre">long</span></tt></li>
<li><tt class="docutils literal"><span class="pre">float</span></tt></li>
<li><tt class="docutils literal"><span class="pre">double</span></tt></li>
</ul>
</blockquote>
<p>In the following descriptions, we reference a generic <tt class="docutils literal"><span class="pre">DATA_TYPE</span></tt>, which
could be any of the C data types listed above.</p>
<div class="section">
<h2><a class="toc-backref" href="#id4" id="input-arrays" name="input-arrays">Input Arrays</a></h2>
<p>Input arrays are defined as arrays of data that are passed into a
routine but are not altered in-place or returned to the user. The
<a class="reference" href="http://www.python.org">python</a> input array is therefore allowed to be almost any <a class="reference" href="http://www.python.org">python</a>
sequence (such as a list) that can be converted to the requested type
of array. The input array signatures are</p>
<blockquote>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">IN_ARRAY1[ANY])</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE*</span> <span class="pre">IN_ARRAY1,</span> <span class="pre">int</span> <span class="pre">DIM1)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(int</span> <span class="pre">DIM1,</span> <span class="pre">DATA_TYPE*</span> <span class="pre">IN_ARRAY1)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">IN_ARRAY2[ANY][ANY])</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE*</span> <span class="pre">IN_ARRAY2,</span> <span class="pre">int</span> <span class="pre">DIM1,</span> <span class="pre">int</span> <span class="pre">DIM2)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(int</span> <span class="pre">DIM1,</span> <span class="pre">int</span> <span class="pre">DIM2,</span> <span class="pre">DATA_TYPE*</span> <span class="pre">IN_ARRAY2)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">IN_ARRAY3[ANY][ANY][ANY])</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE*</span> <span class="pre">IN_ARRAY3,</span> <span class="pre">int</span> <span class="pre">DIM1,</span> <span class="pre">int</span> <span class="pre">DIM2,</span> <span class="pre">int</span> <span class="pre">DIM3)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(int</span> <span class="pre">DIM1,</span> <span class="pre">int</span> <span class="pre">DIM2,</span> <span class="pre">int</span> <span class="pre">DIM3,</span> <span class="pre">DATA_TYPE*</span> <span class="pre">IN_ARRAY3)</span></tt></li>
</ul>
</blockquote>
<p>The first signature listed, <tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">IN_ARRAY[ANY])</span></tt> is for
one-dimensional arrays with hard-coded dimensions. Likewise,
<tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">IN_ARRAY2[ANY][ANY])</span></tt> is for two-dimensional arrays with
hard-coded dimensions, and similarly for three-dimensional.</p>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id5" id="in-place-arrays" name="in-place-arrays">In-Place Arrays</a></h2>
<p>In-place arrays are defined as arrays that are modified in-place. The
input values may or may not be used, but the values at the time the
function returns are significant. The provided <a class="reference" href="http://www.python.org">python</a> argument
must therefore be a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array of the required type. The in-place
signatures are</p>
<blockquote>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">INPLACE_ARRAY1[ANY])</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE*</span> <span class="pre">INPLACE_ARRAY1,</span> <span class="pre">int</span> <span class="pre">DIM1)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(int</span> <span class="pre">DIM1,</span> <span class="pre">DATA_TYPE*</span> <span class="pre">INPLACE_ARRAY1)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">INPLACE_ARRAY2[ANY][ANY])</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE*</span> <span class="pre">INPLACE_ARRAY2,</span> <span class="pre">int</span> <span class="pre">DIM1,</span> <span class="pre">int</span> <span class="pre">DIM2)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(int</span> <span class="pre">DIM1,</span> <span class="pre">int</span> <span class="pre">DIM2,</span> <span class="pre">DATA_TYPE*</span> <span class="pre">INPLACE_ARRAY2)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">INPLACE_ARRAY3[ANY][ANY][ANY])</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE*</span> <span class="pre">INPLACE_ARRAY3,</span> <span class="pre">int</span> <span class="pre">DIM1,</span> <span class="pre">int</span> <span class="pre">DIM2,</span> <span class="pre">int</span> <span class="pre">DIM3)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(int</span> <span class="pre">DIM1,</span> <span class="pre">int</span> <span class="pre">DIM2,</span> <span class="pre">int</span> <span class="pre">DIM3,</span> <span class="pre">DATA_TYPE*</span> <span class="pre">INPLACE_ARRAY3)</span></tt></li>
</ul>
</blockquote>
<p>These typemaps now check to make sure that the <tt class="docutils literal"><span class="pre">INPLACE_ARRAY</span></tt>
arguments use native byte ordering. If not, an exception is raised.</p>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id6" id="argout-arrays" name="argout-arrays">Argout Arrays</a></h2>
<p>Argout arrays are arrays that appear in the input arguments in C, but
are in fact output arrays. This pattern occurs often when there is
more than one output variable and the single return argument is
therefore not sufficient. In <a class="reference" href="http://www.python.org">python</a>, the convential way to return
multiple arguments is to pack them into a sequence (tuple, list, etc.)
and return the sequence. This is what the argout typemaps do. If a
wrapped function that uses these argout typemaps has more than one
return argument, they are packed into a tuple or list, depending on
the version of <a class="reference" href="http://www.python.org">python</a>. The <a class="reference" href="http://www.python.org">python</a> user does not pass these
arrays in, they simply get returned. The argout signatures are</p>
<blockquote>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">ARGOUT_ARRAY1[ANY])</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE*</span> <span class="pre">ARGOUT_ARRAY1,</span> <span class="pre">int</span> <span class="pre">DIM1)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(int</span> <span class="pre">DIM1,</span> <span class="pre">DATA_TYPE*</span> <span class="pre">ARGOUT_ARRAY1)</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">ARGOUT_ARRAY2[ANY][ANY])</span></tt></li>
<li><tt class="docutils literal"><span class="pre">(DATA_TYPE</span> <span class="pre">ARGOUT_ARRAY3[ANY][ANY][ANY])</span></tt></li>
</ul>
</blockquote>
<p>These are typically used in situations where in C/C++, you would
allocate a(n) array(s) on the heap, and call the function to fill the
array(s) values. In <a class="reference" href="http://www.python.org">python</a>, the arrays are allocated for you and
returned as new array objects.</p>
<p>Note that we support <tt class="docutils literal"><span class="pre">DATA_TYPE*</span></tt> argout typemaps in 1D, but not 2D
or 3D. This because of a quirk with the <a class="reference" href="http://www.swig.org">SWIG</a> typemap syntax and
cannot be avoided. Note that for these types of 1D typemaps, the
<a class="reference" href="http://www.python.org">python</a> function will take a single argument representing <tt class="docutils literal"><span class="pre">DIM1</span></tt>.</p>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id7" id="output-arrays" name="output-arrays">Output Arrays</a></h2>
<p>The <tt class="docutils literal"><span class="pre">numpy.i</span></tt> interface file does not support typemaps for output
arrays, for several reasons. First, C/C++ function return arguments
do not have names, so signatures for <tt class="docutils literal"><span class="pre">%typemap(out)</span></tt> do not include
names. This means that if <tt class="docutils literal"><span class="pre">numpy.i</span></tt> supported them, they would
apply to all pointer return arguments for the supported numeric
types. This seems too dangerous. Second, C/C++ return arguments are
limited to a single value. This prevents obtaining dimension
information in a general way. Third, arrays with hard-coded lengths
are not permitted as return arguments. In other words:</p>
<pre class="literal-block">
double[3] newVector(double x, double y, double z);
</pre>
<p>is not legal C/C++ syntax. Therefore, we cannot provide typemaps of
the form:</p>
<pre class="literal-block">
%typemap(out) (TYPE[ANY]);
</pre>
<p>If you run into a situation where a function or method is returning a
pointer to an array, your best bet is to write your own version of the
function to be wrapped, either with <tt class="docutils literal"><span class="pre">%extend</span></tt> for the case of class
methods or <tt class="docutils literal"><span class="pre">%ignore</span></tt> and <tt class="docutils literal"><span class="pre">%rename</span></tt> for the case of functions.</p>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id8" id="other-common-types-bool" name="other-common-types-bool">Other Common Types: bool</a></h2>
<p>Note that C++ type <tt class="docutils literal"><span class="pre">bool</span></tt> is not supported in the list in the
<a class="reference" href="#available-typemaps">Available Typemaps</a> section. NumPy bools are a single byte, while
the C++ <tt class="docutils literal"><span class="pre">bool</span></tt> is four bytes (at least on my system). Therefore:</p>
<pre class="literal-block">
%numpy_typemaps(bool, NPY_BOOL, int)
</pre>
<p>will result in typemaps that will produce code that reference
improper data lengths. You can implement the following macro
expansion:</p>
<pre class="literal-block">
%numpy_typemaps(bool, NPY_UINT, int)
</pre>
<p>to fix the data length problem, and <a class="reference" href="#input-arrays">Input Arrays</a> will work fine,
but <a class="reference" href="#in-place-arrays">In-Place Arrays</a> might fail type-checking.</p>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id9" id="other-common-types-complex" name="other-common-types-complex">Other Common Types: complex</a></h2>
<p>Typemap conversions for complex floating-point types is also not
supported automatically. This is because <a class="reference" href="http://www.python.org">python</a> and <a class="reference" href="http://numpy.scipy.org">NumPy</a> are
written in C, which does not have native complex types. Both
<a class="reference" href="http://www.python.org">python</a> and <a class="reference" href="http://numpy.scipy.org">NumPy</a> implement their own (essentially equivalent)
<tt class="docutils literal"><span class="pre">struct</span></tt> definitions for complex variables:</p>
<pre class="literal-block">
/* Python */
typedef struct {double real; double imag;} Py_complex;
/* NumPy */
typedef struct {float real, imag;} npy_cfloat;
typedef struct {double real, imag;} npy_cdouble;
</pre>
<p>We could have implemented:</p>
<pre class="literal-block">
%numpy_typemaps(Py_complex , NPY_CDOUBLE, int)
%numpy_typemaps(npy_cfloat , NPY_CFLOAT , int)
%numpy_typemaps(npy_cdouble, NPY_CDOUBLE, int)
</pre>
<p>which would have provided automatic type conversions for arrays of
type <tt class="docutils literal"><span class="pre">Py_complex</span></tt>, <tt class="docutils literal"><span class="pre">npy_cfloat</span></tt> and <tt class="docutils literal"><span class="pre">npy_cdouble</span></tt>. However, it
seemed unlikely that there would be any independent (non-<a class="reference" href="http://www.python.org">python</a>,
non-<a class="reference" href="http://numpy.scipy.org">NumPy</a>) application code that people would be using <a class="reference" href="http://www.swig.org">SWIG</a> to
generate a <a class="reference" href="http://www.python.org">python</a> interface to, that also used these definitions
for complex types. More likely, these application codes will define
their own complex types, or in the case of C++, use <tt class="docutils literal"><span class="pre">std::complex</span></tt>.
Assuming these data structures are compatible with <a class="reference" href="http://www.python.org">python</a> and
<a class="reference" href="http://numpy.scipy.org">NumPy</a> complex types, <tt class="docutils literal"><span class="pre">%numpy_typemap</span></tt> expansions as above (with
the user's complex type substituted for the first argument) should
work.</p>
</div>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id10" id="helper-functions" name="helper-functions">Helper Functions</a></h1>
<p>The <tt class="docutils literal"><span class="pre">numpy.i</span></tt> file containes several macros and routines that it
uses internally to build its typemaps. However, these functions may
be useful elsewhere in your interface file.</p>
<div class="section">
<h2><a class="toc-backref" href="#id11" id="macros" name="macros">Macros</a></h2>
<blockquote>
<dl class="docutils">
<dt><strong>is_array(a)</strong></dt>
<dd>Evaluates as true if <tt class="docutils literal"><span class="pre">a</span></tt> is non-<tt class="docutils literal"><span class="pre">NULL</span></tt> and can be cast to a
<tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt>.</dd>
<dt><strong>array_type(a)</strong></dt>
<dd>Evaluates to the integer data type code of <tt class="docutils literal"><span class="pre">a</span></tt>, assuming <tt class="docutils literal"><span class="pre">a</span></tt> can
be cast to a <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt>.</dd>
<dt><strong>array_numdims(a)</strong></dt>
<dd>Evaluates to the integer number of dimensions of <tt class="docutils literal"><span class="pre">a</span></tt>, assuming
<tt class="docutils literal"><span class="pre">a</span></tt> can be cast to a <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt>.</dd>
<dt><strong>array_dimensions(a)</strong></dt>
<dd>Evaluates to an array of type <tt class="docutils literal"><span class="pre">npy_intp</span></tt> and length
<tt class="docutils literal"><span class="pre">array_numdims(a)</span></tt>, giving the lengths of all of the dimensions
of <tt class="docutils literal"><span class="pre">a</span></tt>, assuming <tt class="docutils literal"><span class="pre">a</span></tt> can be cast to a <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt>.</dd>
<dt><strong>array_size(a,i)</strong></dt>
<dd>Evaluates to the <tt class="docutils literal"><span class="pre">i</span></tt>-th dimension size of <tt class="docutils literal"><span class="pre">a</span></tt>, assuming <tt class="docutils literal"><span class="pre">a</span></tt>
can be cast to a <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt>.</dd>
<dt><strong>array_data(a)</strong></dt>
<dd>Evaluates to a pointer of type <tt class="docutils literal"><span class="pre">void*</span></tt> that points to the data
buffer of <tt class="docutils literal"><span class="pre">a</span></tt>, assuming <tt class="docutils literal"><span class="pre">a</span></tt> can be cast to a <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt>.</dd>
<dt><strong>array_is_contiguous(a)</strong></dt>
<dd>Evaluates as true if <tt class="docutils literal"><span class="pre">a</span></tt> is a contiguous array. Equivalent to
<tt class="docutils literal"><span class="pre">(PyArray_ISCONTIGUOUS(a))</span></tt>.</dd>
<dt><strong>array_is_native(a)</strong></dt>
<dd>Evaluates as true if the data buffer of <tt class="docutils literal"><span class="pre">a</span></tt> uses native byte
order. Equivalent to <tt class="docutils literal"><span class="pre">(PyArray_ISNOTSWAPPED(a))</span></tt>.</dd>
</dl>
</blockquote>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id12" id="routines" name="routines">Routines</a></h2>
<blockquote>
<p><strong>pytype_string()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">char*</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyObject*</span> <span class="pre">py_obj</span></tt>, a general <a class="reference" href="http://www.python.org">python</a> object.</li>
</ul>
<p>Return a string describing the type of <tt class="docutils literal"><span class="pre">py_obj</span></tt>.</p>
</blockquote>
<p><strong>typecode_string()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">char*</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">typecode</span></tt>, a <a class="reference" href="http://numpy.scipy.org">NumPy</a> integer typecode.</li>
</ul>
<p>Return a string describing the type corresponding to the <a class="reference" href="http://numpy.scipy.org">NumPy</a>
<tt class="docutils literal"><span class="pre">typecode</span></tt>.</p>
</blockquote>
<p><strong>type_match()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">int</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">actual_type</span></tt>, the <a class="reference" href="http://numpy.scipy.org">NumPy</a> typecode of a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">desired_type</span></tt>, the desired <a class="reference" href="http://numpy.scipy.org">NumPy</a> typecode.</li>
</ul>
<p>Make sure that <tt class="docutils literal"><span class="pre">actual_type</span></tt> is compatible with
<tt class="docutils literal"><span class="pre">desired_type</span></tt>. For example, this allows character and
byte types, or int and long types, to match. This is now
equivalent to <tt class="docutils literal"><span class="pre">PyArray_EquivTypenums()</span></tt>.</p>
</blockquote>
<p><strong>obj_to_array_no_conversion()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyObject*</span> <span class="pre">input</span></tt>, a general <a class="reference" href="http://www.python.org">python</a> object.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">typecode</span></tt>, the desired <a class="reference" href="http://numpy.scipy.org">NumPy</a> typecode.</li>
</ul>
<p>Cast <tt class="docutils literal"><span class="pre">input</span></tt> to a <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt> if legal, and ensure that
it is of type <tt class="docutils literal"><span class="pre">typecode</span></tt>. If <tt class="docutils literal"><span class="pre">input</span></tt> cannot be cast, or the
<tt class="docutils literal"><span class="pre">typecode</span></tt> is wrong, set a <a class="reference" href="http://www.python.org">python</a> error and return <tt class="docutils literal"><span class="pre">NULL</span></tt>.</p>
</blockquote>
<p><strong>obj_to_array_allow_conversion()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyObject*</span> <span class="pre">input</span></tt>, a general <a class="reference" href="http://www.python.org">python</a> object.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">typecode</span></tt>, the desired <a class="reference" href="http://numpy.scipy.org">NumPy</a> typecode of the resulting
array.</li>
<li><tt class="docutils literal"><span class="pre">int*</span> <span class="pre">is_new_object</span></tt>, returns a value of 0 if no conversion
performed, else 1.</li>
</ul>
<p>Convert <tt class="docutils literal"><span class="pre">input</span></tt> to a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array with the given <tt class="docutils literal"><span class="pre">typecode</span></tt>.
On success, return a valid <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt> with the correct
type. On failure, the <a class="reference" href="http://www.python.org">python</a> error string will be set and the
routine returns <tt class="docutils literal"><span class="pre">NULL</span></tt>.</p>
</blockquote>
<p><strong>make_contiguous()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyArrayObject*</span> <span class="pre">ary</span></tt>, a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array.</li>
<li><tt class="docutils literal"><span class="pre">int*</span> <span class="pre">is_new_object</span></tt>, returns a value of 0 if no conversion
performed, else 1.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">min_dims</span></tt>, minimum allowable dimensions.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">max_dims</span></tt>, maximum allowable dimensions.</li>
</ul>
<p>Check to see if <tt class="docutils literal"><span class="pre">ary</span></tt> is contiguous. If so, return the input
pointer and flag it as not a new object. If it is not contiguous,
create a new <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt> using the original data, flag it
as a new object and return the pointer.</p>
</blockquote>
<p><strong>obj_to_array_contiguous_allow_conversion()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyObject*</span> <span class="pre">input</span></tt>, a general <a class="reference" href="http://www.python.org">python</a> object.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">typecode</span></tt>, the desired <a class="reference" href="http://numpy.scipy.org">NumPy</a> typecode of the resulting
array.</li>
<li><tt class="docutils literal"><span class="pre">int*</span> <span class="pre">is_new_object</span></tt>, returns a value of 0 if no conversion
performed, else 1.</li>
</ul>
<p>Convert <tt class="docutils literal"><span class="pre">input</span></tt> to a contiguous <tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt> of the
specified type. If the input object is not a contiguous
<tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt>, a new one will be created and the new object
flag will be set.</p>
</blockquote>
<p><strong>require_contiguous()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">int</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyArrayObject*</span> <span class="pre">ary</span></tt>, a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array.</li>
</ul>
<p>Test whether <tt class="docutils literal"><span class="pre">ary</span></tt> is contiguous. If so, return 1. Otherwise,
set a <a class="reference" href="http://www.python.org">python</a> error and return 0.</p>
</blockquote>
<p><strong>require_native()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">int</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyArray_Object*</span> <span class="pre">ary</span></tt>, a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array.</li>
</ul>
<p>Require that <tt class="docutils literal"><span class="pre">ary</span></tt> is not byte-swapped. If the array is not
byte-swapped, return 1. Otherwise, set a <a class="reference" href="http://www.python.org">python</a> error and
return 0.</p>
</blockquote>
<p><strong>require_dimensions()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">int</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyArrayObject*</span> <span class="pre">ary</span></tt>, a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">exact_dimensions</span></tt>, the desired number of dimensions.</li>
</ul>
<p>Require <tt class="docutils literal"><span class="pre">ary</span></tt> to have a specified number of dimensions. If the
array has the specified number of dimensions, return 1.
Otherwise, set a <a class="reference" href="http://www.python.org">python</a> error and return 0.</p>
</blockquote>
<p><strong>require_dimensions_n()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">int</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyArrayObject*</span> <span class="pre">ary</span></tt>, a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array.</li>
<li><tt class="docutils literal"><span class="pre">int*</span> <span class="pre">exact_dimensions</span></tt>, an array of integers representing
acceptable numbers of dimensions.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">n</span></tt>, the length of <tt class="docutils literal"><span class="pre">exact_dimensions</span></tt>.</li>
</ul>
<p>Require <tt class="docutils literal"><span class="pre">ary</span></tt> to have one of a list of specified number of
dimensions. If the array has one of the specified number of
dimensions, return 1. Otherwise, set the <a class="reference" href="http://www.python.org">python</a> error string
and return 0.</p>
</blockquote>
<p><strong>require_size()</strong></p>
<blockquote>
<p>Return type: <tt class="docutils literal"><span class="pre">int</span></tt></p>
<p>Arguments:</p>
<ul class="simple">
<li><tt class="docutils literal"><span class="pre">PyArrayObject*</span> <span class="pre">ary</span></tt>, a <a class="reference" href="http://numpy.scipy.org">NumPy</a> array.</li>
<li><tt class="docutils literal"><span class="pre">npy_int*</span> <span class="pre">size</span></tt>, an array representing the desired lengths of
each dimension.</li>
<li><tt class="docutils literal"><span class="pre">int</span> <span class="pre">n</span></tt>, the length of <tt class="docutils literal"><span class="pre">size</span></tt>.</li>
</ul>
<p>Require <tt class="docutils literal"><span class="pre">ary</span></tt> to have a specified shape. If the array has the
specified shape, return 1. Otherwise, set the <a class="reference" href="http://www.python.org">python</a> error
string and return 0.</p>
</blockquote>
</blockquote>
</div>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id13" id="beyond-the-provided-typemaps" name="beyond-the-provided-typemaps">Beyond the Provided Typemaps</a></h1>
<p>There are many C or C++ array/<a class="reference" href="http://numpy.scipy.org">NumPy</a> array situations not covered by
a simple <tt class="docutils literal"><span class="pre">%include</span> <span class="pre">"numpy.i"</span></tt> and subsequent <tt class="docutils literal"><span class="pre">%apply</span></tt> directives.</p>
<div class="section">
<h2><a class="toc-backref" href="#id14" id="a-common-example" name="a-common-example">A Common Example</a></h2>
<p>Consider a reasonable prototype for a dot product function:</p>
<pre class="literal-block">
double dot(int len, double* vec1, double* vec2);
</pre>
<p>The <a class="reference" href="http://www.python.org">python</a> interface that we want is:</p>
<pre class="literal-block">
def dot(vec1, vec2):
"""
dot(PyObject,PyObject) -> double
"""
</pre>
<p>The problem here is that there is one dimension argument and two array
arguments, and our typemaps are set up for dimensions that apply to a
single array (in fact, <a class="reference" href="http://www.swig.org">SWIG</a> does not provide a mechanism for
associating <tt class="docutils literal"><span class="pre">len</span></tt> with <tt class="docutils literal"><span class="pre">vec2</span></tt> that takes two <a class="reference" href="http://www.python.org">python</a> input
arguments). The recommended solution is the following:</p>
<pre class="literal-block">
%apply (int DIM1, double* IN_ARRAY1) {(int len1, double* vec1),
(int len2, double* vec2)}
%rename (dot) my_dot;
%inline %{
double my_dot(int len1, double* vec1, int len2, double* vec2) {
if (len1 != len2) {
PyErr_Format(PyExc_ValueError,
"Arrays of lengths (%d,%d) given",
len1, len2);
return 0.0;
}
return dot(len1, vec1, vec2);
}
%}
</pre>
<p>If the header file that contains the prototype for <tt class="docutils literal"><span class="pre">double</span> <span class="pre">dot()</span></tt>
also contains other prototypes that you want to wrap, so that you need
to <tt class="docutils literal"><span class="pre">%include</span></tt> this header file, then you will also need a <tt class="docutils literal"><span class="pre">%ignore</span>
<span class="pre">dot;</span></tt> directive, placed after the <tt class="docutils literal"><span class="pre">%rename</span></tt> and before the
<tt class="docutils literal"><span class="pre">%include</span></tt> directives. Or, if the function in question is a class
method, you will want to use <tt class="docutils literal"><span class="pre">%extend</span></tt> rather than <tt class="docutils literal"><span class="pre">%inline</span></tt> in
addition to <tt class="docutils literal"><span class="pre">%ignore</span></tt>.</p>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id15" id="other-situations" name="other-situations">Other Situations</a></h2>
<p>There are other wrapping situations in which <tt class="docutils literal"><span class="pre">numpy.i</span></tt> may be
helpful when you encounter them.</p>
<blockquote>
<ul>
<li><p class="first">In some situations, it is possible that you could use the
<tt class="docutils literal"><span class="pre">%numpy_templates</span></tt> macro to implement typemaps for your own
types. See the <a class="reference" href="#other-common-types-bool">Other Common Types: bool</a> or <a class="reference" href="#other-common-types-complex">Other Common
Types: complex</a> sections for examples. Another situation is if
your dimensions are of a type other than <tt class="docutils literal"><span class="pre">int</span></tt> (say <tt class="docutils literal"><span class="pre">long</span></tt> for
example):</p>
<pre class="literal-block">
%numpy_typemaps(double, NPY_DOUBLE, long)
</pre>
</li>
<li><p class="first">You can use the code in <tt class="docutils literal"><span class="pre">numpy.i</span></tt> to write your own typemaps.
For example, if you had a four-dimensional array as a function
argument, you could cut-and-paste the appropriate
three-dimensional typemaps into your interface file. The
modifications for the fourth dimension would be trivial.</p>
</li>
<li><p class="first">Sometimes, the best approach is to use the <tt class="docutils literal"><span class="pre">%extend</span></tt> directive
to define new methods for your classes (or overload existing ones)
that take a <tt class="docutils literal"><span class="pre">PyObject*</span></tt> (that either is or can be converted to a
<tt class="docutils literal"><span class="pre">PyArrayObject*</span></tt>) instead of a pointer to a buffer. In this
case, the helper routines in <tt class="docutils literal"><span class="pre">numpy.i</span></tt> can be very useful.</p>
</li>
<li><p class="first">Writing typemaps can be a bit nonintuitive. If you have specific
questions about writing <a class="reference" href="http://www.swig.org">SWIG</a> typemaps for <a class="reference" href="http://numpy.scipy.org">NumPy</a>, the
developers of <tt class="docutils literal"><span class="pre">numpy.i</span></tt> do monitor the
<a class="reference" href="mailto:Numpy-discussion@scipy.org">Numpy-discussion</a> and
<a class="reference" href="mailto:Swig-user@lists.sourceforge.net">Swig-user</a> mail lists.</p>
</li>
</ul>
</blockquote>
</div>
<div class="section">
<h2><a class="toc-backref" href="#id16" id="a-final-note" name="a-final-note">A Final Note</a></h2>
<p>When you use the <tt class="docutils literal"><span class="pre">%apply</span></tt> directive, as is usually necessary to use
<tt class="docutils literal"><span class="pre">numpy.i</span></tt>, it will remain in effect until you tell <a class="reference" href="http://www.swig.org">SWIG</a> that it
shouldn't be. If the arguments to the functions or methods that you
are wrapping have common names, such as <tt class="docutils literal"><span class="pre">length</span></tt> or <tt class="docutils literal"><span class="pre">vector</span></tt>,
these typemaps may get applied in situations you do not expect or
want. Therefore, it is always a good idea to add a <tt class="docutils literal"><span class="pre">%clear</span></tt>
directive after you are done with a specific typemap:</p>
<pre class="literal-block">
%apply (double* IN_ARRAY1, int DIM1) {(double* vector, int length)}
%include "my_header.h"
%clear (double* vector, int length);
</pre>
<p>In general, you should target these typemap signatures specifically
where you want them, and then clear them after you are done.</p>
</div>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id17" id="summary" name="summary">Summary</a></h1>
<p>Out of the box, <tt class="docutils literal"><span class="pre">numpy.i</span></tt> provides typemaps that support conversion
between <a class="reference" href="http://numpy.scipy.org">NumPy</a> arrays and C arrays:</p>
<blockquote>
<ul class="simple">
<li>That can be one of 12 different scalar types: <tt class="docutils literal"><span class="pre">signed</span> <span class="pre">char</span></tt>,
<tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">char</span></tt>, <tt class="docutils literal"><span class="pre">short</span></tt>, <tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">short</span></tt>, <tt class="docutils literal"><span class="pre">int</span></tt>,
<tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">int</span></tt>, <tt class="docutils literal"><span class="pre">long</span></tt>, <tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">long</span></tt>, <tt class="docutils literal"><span class="pre">long</span> <span class="pre">long</span></tt>,
<tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">long</span> <span class="pre">long</span></tt>, <tt class="docutils literal"><span class="pre">float</span></tt> and <tt class="docutils literal"><span class="pre">double</span></tt>.</li>
<li>That support 23 different argument signatures for each data type,
including:<ul>
<li>One-dimensional, two-dimensional and three-dimensional arrays.</li>
<li>Input-only, in-place, and argout behavior.</li>
<li>Hard-coded dimensions, data-buffer-then-dimensions
specification, and dimensions-then-data-buffer specification.</li>
</ul>
</li>
</ul>
</blockquote>
<p>The <tt class="docutils literal"><span class="pre">numpy.i</span></tt> interface file also provides additional tools for
wrapper developers, including:</p>
<blockquote>
<ul class="simple">
<li>A <a class="reference" href="http://www.swig.org">SWIG</a> macro (<tt class="docutils literal"><span class="pre">%numpy_typemaps</span></tt>) with three arguments for
implementing the 23 argument signatures for the user's choice of
(1) C data type, (2) <a class="reference" href="http://numpy.scipy.org">NumPy</a> data type (assuming they match), and
(3) dimension type.</li>
<li>Seven C macros and eleven C functions that can be used to write
specialized typemaps, extensions, or inlined functions that handle
cases not covered by the provided typemaps.</li>
</ul>
</blockquote>
</div>
<div class="section">
<h1><a class="toc-backref" href="#id18" id="acknowledgements" name="acknowledgements">Acknowledgements</a></h1>
<p>Many people have worked to glue <a class="reference" href="http://www.swig.org">SWIG</a> and <a class="reference" href="http://numpy.scipy.org">NumPy</a> together (as well
as <a class="reference" href="http://www.swig.org">SWIG</a> and the predecessors of <a class="reference" href="http://numpy.scipy.org">NumPy</a>, Numeric and numarray).
The effort to standardize this work into <tt class="docutils literal"><span class="pre">numpy.i</span></tt> began at the 2005
<a class="reference" href="http://scipy.org">SciPy</a> Conference with a conversation between
Fernando Perez and myself. Fernando collected helper functions and
typemaps from Michael Hunter, Anna Omelchenko and Michael Sanner.
Sebastian Hasse has also provided additional error checking and use
cases. The work of these contributors has made this end result
possible.</p>
</div>
</div>
<div class="footer">
<hr class="footer" />
Generated on: 2007-04-13 20:43 UTC.
Generated by <a class="reference" href="http://docutils.sourceforge.net/">Docutils</a> from <a class="reference" href="http://docutils.sourceforge.net/rst.html">reStructuredText</a> source.
</div>
</body>
</html>
|
